Hi Taeyun,
Thanks for the writeup! Let me reflect to some areas:
the caller manages the version identifier separately.
Since the callers of this interface would be the query engines
themselves most of the cases, this would mean that Impala, Spark, Hive,
Trino, etc. would need to implement their way of storing and updating
VersionIdentifiers. This would push unwanted complexity to the client side.
I'm against this. I'd be ok to include this VersionIdentifier (or
CatalogVersion, or doesn't matter how we called this) into the Table object
and sort of hide it from the clients if the community allows us to do so.
caller must rely on exception handling
I don't think this approach is new. Even the current loadTable() API throws
an exception if the table doesn't exist for instance. We can similarly
throw an exception if the freshness-aware table loading is not feasible for
some reason.
I'm against ... where the function always loads and returns a new Table
> when freshness checks aren’t possible
We are in agreement here, I'm also against this. The user expects better
performance when calling this interface compared to keeping calling the
regular loadTable(). So if freshness checks aren't possible, let's say
because that catalog implementation doesn't support it, the user should get
an exception. I don't think that there is a need to call
canCheckFreshness() before. That's just extra noise in the interface.
The server is free to assign any value to the ETag. This means the client
> should not attempt to interpret the content of the ETag.
With the REST spec we can have a control of what the implementations should
put into this ETag header. If we articulate in the spec that the ETag
should be the metadata location then the implementations have to follow
this contract, hence the clients could give semantics to the ETag and use
it for the metadata location.
I wonder if you have any proposal for the content of the ETag apart from it
can be any value. Another approach that comes to my mind is to create a
LoadTableResponse object on the server side, hash it, and then the hash of
the response body could be used as an ETag. However, for that the server
has to construct the LoadTableResponse unconditionally, even though in some
cases this won't be sent back to the client. With this the server has to
read and parse the full table metadata in order to judge if the table has
changed or not. This means that there is going to be the same load on the
server as if the user were calling the regular loadTable() interface and
just the amount of data sent back on the network would be less.
Hence I propose that even though in theory the REST servers could use
anything as an ETag, the metadata location seems a pretty convenient
content for that header, and then we don't have to do a full table metadata
read on the server side, just get the metadata location that is most
probably cached anyway in memory.
*Next steps*:
I think our proposals are coming close to each other, even though we have
some disagreements on some details. I feel that we have really low activity
on this thread from people with decision making privileges so let's do this:
Let me put together an improvement proposal as advised by Fokko, and let's
hope it will attract people having binding votes so that we can come to a
conclusion on all details. What do you think?
Regards,
Gabor
On Fri, Nov 22, 2024 at 3:06 AM Taeyun Kim <taeyun....@innowireless.com>
wrote:
> Hi,
>
> - On the Function:
>
> The function signature I propose is as follows (slightly modified from my
> previous suggestion):
>
> Option(Table, Option(VersionIdentifier))
> loadTableIfChanged(TableIdentifier, Option(VersionIdentifier))
>
> The key difference from Gabor’s proposal is that the caller manages the
> version identifier separately. For example, in the case of a REST catalog,
> the VersionIdentifier could be an ETag, while in other catalogs, it could
> be a metadata location. The important point is that the caller doesn’t
> interpret the identifier. In Java, the type for the version identifier
> could even be Object, allowing it to represent the Table itself if
> necessary.
> With this signature, the callee can return None as the VersionIdentifier
> to signal that freshness checks are not possible for the table, effectively
> informing the caller that a reload is unavoidable. (In Java, this could be
> implemented with Optional or simply using a nullable value as before.)
> In Gabor’s proposal, the caller must rely on exception handling to
> determine if freshness checks are possible. However, exceptions are
> typically used for unexpected issues, so using them for this purpose might
> feel slightly awkward.
> That said, as long as there’s a way for the caller to know whether
> freshness checks are possible, the exact function signature might not be
> critically important.
> Another key point is that the caller provides the basis for the freshness
> check when invoking the function. Different callers might hold distinct
> versions, so it’s important for the caller to supply the version
> information. In my proposal, the caller provides this explicitly via the
> VersionIdentifier. In Gabor’s proposal, version information would likely
> need to be embedded within the Table object.
> For Gabor’s approach, adding a non-static method to the Table class like
> the following could allow the caller to pre-check freshness capabilities:
>
> bool Table.canCheckFreshness()
>
> On the other hand, I’m against an implementation of loadTableIfChanged()
> where the function always loads and returns a new Table when freshness
> checks aren’t possible. This approach prevents the caller from handling
> caching behavior separately based on the availability of freshness checks.
> This is similar to Gabor’s concern about losing control of caching to the
> 4) and 5) layers.
> Therefore, if freshness checks are not possible, I agree with Gabor that
> the function (at least in Java) should throw an
> UnsupportedOperationException.
> While freshness checks are relatively straightforward to implement (e.g.,
> by comparing metadata locations), and all catalogs may eventually support
> this feature soon after the API is introduced, it’s not a guarantee. For
> example, a RESTClient library class that supports freshness checks might be
> used to connect to an older REST catalog server that doesn’t support the
> new freshness checking specification. The client may not have the authority
> to upgrade the server. BTW, the RESTClient can determine that the server
> doesn’t support freshness checks based on the absence of these HTTP caching
> headers.
>
> - On ETag Content:
>
> The server is free to assign any value to the ETag. This means the client
> should not attempt to interpret the content of the ETag.
> As I mentioned before, if the REST catalog API uses ETags, it’s essential
> that no semantic meaning is attributed to their values between client and
> server.
>
> Thank you.
>
>
> -----Original Message-----
> From: "Gabor Kaszab" <gaborkas...@apache.org>
> To: <dev@iceberg.apache.org>;
> Cc:
> Sent: 2024-11-21 (목) 21:06:48 (UTC+09:00)
> Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the latest
>
>
> Hey,
>
> I think there is one open question here where we disagree: It's the
> proposed function on the Catalog API (not the REST spec). I don't think we
> can ever include a parameter like ETag at this level of abstraction. The
> Catalog API is common for all the catalog implementations and is not just
> for REST or for catalogs that use HTTP. Since ETag is an HTTP specific
> detail, hence I said it's an implementation detail and we can't include it
> to the Catalog level API. It is relevant for the proposed REST spec changes
> and the HTTPClient implementation within the REST client, but it is most
> probably not relevant for other catalog types like HiveCatalog,
> HadoopCatalog, etc.
>
>
> In terms of the Catalog level API (levels 2) and 3) in my previous mail) I
> think this new API should be used only for freshness aware table loading
> and we shouldn't fall back to regular table loading if this is not
> implemented by a catalog. I find the other naming more verbose for this
> purpose:
>
>
> Table loadTableIfChanged(TableIdentifier, Table);
>
>
> Here, the Table parameter could be null if the client loads it the first
> time, hence the need for a TableIdentifier parameter. If a catalog doesn't
> have a freshness aware table loading mechanism, it would throw an
> UnsupportedOperationException. This would avoid confusion whether each call
> on this API is freshness aware or not.
>
>
> Note, there is no ETag included here explicitly in the function signature.
> - Initially this is how I thought this should work:
> The level of abstraction takes care of ETags where they are relevant, so
> in our case it's the REST client (most probably level 3) in my prev mail).
> This level could have a mapping between TableIdentifiers and latest known
> ETags. So when a freshness aware table loading request comes to the REST
> client it looks up the ETag using the TableIdentifier and uses it for the
> HTTP header.
> After talking with Zoltan we found some risks with this approach in
> concurrent scenarios, where the first thread gets the latest table, sets
> the latest ETag within this mapping, but then other threads still holding
> an old version of the table could get stuck with this old version since the
> REST client's ETag is at the latest so no actual reload would be performed.
>
> - Alternatives:
> 1) The Table class, or in fact the inherited classes should have a field
> that stores the 'catalogVersion' as suggested by Zoltan. For REST catalogs
> this can store the ETag, for other catalogs some other information for the
> same purpose.
> 2) I checked this description of ETags, and even though we discussed
> earlier that this is some server generated information, for me it seems
> that it can be basically anything:
> "There are no restrictions on how the server must generate the value, so
> servers are free to set the value based on whatever means they choose —
> such as a hash of the body contents or a version number." So basically an
> ETag can also be a metadata location String as suggested by Yufei (if I'm
> not mistaken). We can also go with this approach and then there is no need
> for a new field within the Table class.
>
>
> Looking forward to hearing your thoughts!
> Gabor
>
>
> On Thu, Nov 21, 2024 at 10:03 AM Zoltán Borók-Nagy <borokna...@apache.org>
> wrote:
>
> Sorry, one more thing about the methods:
>
> Table reloadTable(Table); // or,
> Table reloadTable(TableIdentifier, Table) // where Table could be NULL
>
> I want to highlight that it is super easy to provide a default
> implementation which just loads the table. Then later, catalog
> implementations can just add their clever tricks to make it more
> efficient.
>
> Cheers,
> Zoltan
>
> On Thu, Nov 21, 2024 at 9:53 AM Zoltán Borók-Nagy <borokna...@apache.org>
> wrote:
> >
> > Hi,
> >
> > I agree with Gabor that the support of efficiently reloading Iceberg
> > tables is a generic problem that applies to all catalog
> > implementations.
> > I also think that the programming API, especially the Iceberg Java
> > library is very important, as almost all Iceberg clients use this
> > library to interact with Iceberg tables, no matter which catalog they
> > reside in.
> > Even the engines that are mostly written in C++ (Impala, Starrocks,
> > Doris) interact with Iceberg tables from their Java frontends using
> > the Iceberg Java library.
> >
> > "Since libraries are open-source, I can modify them as needed for my
> > use case" - if you want to maintain a private fork, then sure,
> > otherwise you really want to avoid introducing breaking changes. Also,
> > you want to introduce new features in a way that is acceptable for the
> > community. In that sense, modifying a library's interface is not much
> > easier than modifying a server's interface. Of course, clients of a
> > library have control over when to upgrade, a privilege you don't
> > always have for server APIs, but this is why API versioning was
> > invented, anyway, we are diverging from the main topic here.
> >
> > Since this reloadTable() method could be useful for other Catalog
> > implementations as well, I think we would like to add a new method to
> > org.apache.iceberg.catalog.Catalog that doesn't take any
> > implementation-specific detail about the underlying catalog. To
> > overcome this, catalogs could embed catalog-specific information into
> > the Table object when they initially load the table, e.g. "String
> > catalogVersion". In the case of the REST Catalog the catalogVersion
> > string would be the ETag. Other catalogs might not even need to add
> > anything, as the metadata_location of the Table object is sufficient.
> >
> > This way the API would be simple and generic:
> >
> > Table reloadTable(Table); // or,
> > Table reloadTable(TableIdentifier, Table) // where Table could be NULL
> >
> > Cheers,
> > Zoltan
> >
> > On Thu, Nov 21, 2024 at 1:51 AM Taeyun Kim <taeyun....@innowireless.com>
> wrote:
> > >
> > > Hi Gabor,
> > >
> > > On HTTP Caching:
> > >
> > > If an HTTP client library performs caching by default and doesn’t
> allow disabling it, I believe that library shouldn’t be used - at least in
> the context of this discussion.
> > > The kind of HTTP client library I have in mind is one that handles
> encoding and decoding of HTTP headers and body, as well as connection
> pooling. The responsibility for interpreting headers, status, and body
> content should remain with the application. While caching support can be
> provided, it should be optional.
> > > When using a library that behaves as I described, the issues you
> mentioned in points 4) and 5) shouldn’t arise, as the library wouldn’t
> interfere with caching.
> > > For reference, the Rust reqwest crate (which Iceberg-Rust appears to
> use) seems to operate as expected in this regard.
> > >
> > > On Programming Languages and APIs:
> > >
> > > One of my points was that there doesn’t seem to be a reason to center
> the discussion around Java (and its libraries).
> > > BTW, I don’t think it’s necessary for the functions in the
> iceberg-rust library to be identical to those in the Java library. Optimal
> solutions may vary by language, and library developers may have different
> design goals.
> > > Personally, my primary focus is on the REST catalog API specification,
> rather than language-specific library APIs. (To avoid confusion, I’ll refer
> to the REST catalog API as the "specification" from here on.)
> > > Library APIs are (merely) implementations designed to make the
> specification easier to use. Since libraries are open-source, I can modify
> them as needed for my use case (and, in fact, I’ve made modifications to
> iceberg-rust for my purposes). However, the specification defines the
> interface between different applications or servers, making it immutable
> for practical purposes.
> > >
> > > On ETags:
> > >
> > > The decision to use ETags is not just an implementation detail - it is
> part of the specification itself. In my view, it is far more significant
> than the signature of a library API function. I’ve outlined the reasons for
> this above.
> > >
> > > On the Proposal:
> > >
> > > I agree that the current function (loadTable(TableIdentifier)) cannot
> be freshness-aware. This is expected, as the caller doesn’t provide the
> version it holds, leaving the callee with no basis for comparison.
> > > On the other hand, the proposed new function signature doesn’t seem to
> provide a way for the caller to supply ETags (or equivalent identifiers
> representing specific table versions for other catalog types). Is such
> information intended to be embedded within the Table structure?
> > > To me, it seems clearer to explicitly provide such information (like
> ETags) rather than embedding it in the Table structure. That said, I might
> be misunderstanding the intention here.
> > >
> > > Thank you.
> > >
> > >
> > > -----Original Message-----
> > > From: "Gabor Kaszab" <gaborkas...@apache.org>
> > > To: <dev@iceberg.apache.org>;
> > > Cc:
> > > Sent: 2024-11-19 (화) 21:26:01 (UTC+09:00)
> > > Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the
> latest
> > >
> > >
> > > Hi,
> > >
> > >
> > > Thanks for sharing your view, Taeyun! I think there are many levels of
> representation here and we might not mean the same with our points. I think
> in general an interaction between a query engine and an Iceberg REST
> catalog has these different layers:
> > > 1) The engine (Impala, Spark, Trino, etc.).
> > > 2) Catalog API of the Iceberg lib offers loadTable(TableIdentifier)
> that returns a Table object. Different language implementations seem to
> have the same API (Java, Rust, etc.).
> > > 3) The particular implementation of a catalog that implements the
> above loadTable(TableIdentifier) function. In this example the RESTCatalog
> / RESTSessionCatalog.
> > > 4) RESTClient implemented by HTTPClient (used by the REST catalog) to
> communicate with the REST server (still implemented within Iceberg)
> > > 5) The external HTTPClient
> (org.apache.hc.client5.http.impl.classic.CloseableHttpClient) that
> orchestrates the HTTP traffic between the client and the server
> > >
> > >
> > > Let me reflect on your comments:
> > > - HTTP caching
> > > With the above layers in mind if I'm not mistaken HTTP Caching would
> be configured in 4) and the actual caching of HTTP responses would be in
> 5). This is what I meant by HTTP layer. With HTTP Caching the control of
> how long a cached TableMetadata is stored will no longer be in 1) and would
> be in 4) - 5). I don't think that any of the engines that cache table
> metadata would want to have this loss of control.
> > >
> > >
> > > - Programming language
> > > I'm not sure I get your point with this. The Catalog API seems the
> same regardless of programming language (See the links for 2) ).
> > >
> > >
> > > - ETags
> > > An ETag is an implementation detail that is relevant for HTTP
> communication. We can't extend the Catalog API in 2) nor in 3) with
> functions that are aware of ETags (e.g. return ETags or accept ETags as
> param). Those APIs are common for all the Catalog implementations including
> ones that don't leverage ETags for HTTP traffic.
> > >
> > >
> > > Proposal:
> > > - Catalog API
> > > I don't think that the current Catalog.loadTable(TableIdentifier) API
> in 2) is suitable for a freshness-aware table loading use case. It wouldn't
> be transparent to the clients if that actual catalog implementation avoids
> reloading the table if it hasn't changed or if that catalog implementation
> reloads the table unconditionally with this API call.
> > > Also it doesn't seem straightforward what the API should return if the
> table is considered fresh. This API returns a Table object and in case we
> get a 304 without a body from the catalog server, we won't have a way to
> construct a Table object as a return value for this API. I already shared
> my concerns for caching the LoadTableResponses within 4) - 5)
> > >
> > >
> > > So I think we need a new API on the Catalog for this purpose. Thanks
> Zoltan for the naming suggestion, after I sent my mail yesterday I also
> thought that I could've come up with a more intuitive name.
> > > This can either be:
> > > a) Table reloadTable(TableIdentifier, Table)
> > > b) Table loadTableIfChanged(TableIdentifier, Table)
> > >
> > >
> > > With this Catalog level API we could provide the current known state
> of that particular table as a parameter, and if the client side of the
> catalog implementation finds that the table hasn't changed it can return
> this Table object for the current state. With this approach no caching
> would be needed in 2) - 5). It's up to the catalog implementation how it
> finds out if the table has been changed or not.
> > >
> > >
> > > - REST API, REST spec
> > > The REST API could use the ETag approach to check table freshness. As
> described in previous mails this could reduce the number of round trips to
> refresh a table to one without the need of separately checking the
> freshness. We could use the same endpoint as we do for the current
> loadTable(), with an additional mention of an optional ETag being used and
> also including the 304 into the possible responses.
> > >
> > >
> > > For this approach we have to cache the [TableIdentifier -> last ETag]
> mapping somewhere. I think 4), the Iceberg specific HTTPClient could be
> suitable for this purpose, however, this seems too low level for this
> purpose. We can also consider RESTSessionCatalog to cache the ETags.
> > >
> > >
> > > This is something to be considered, but for REST catalog
> implementations that don't support the ETag based implementation, they
> would just simply perform a regular loadTable operation, not bothering with
> sending 304 codes. We can also investigate if we should get an exception if
> that particular REST implementation doesn't support the ETag approach, so
> that clients could notice if there is no freshness-aware table loading
> under the hood.
> > >
> > >
> > > - Other catalog types
> > > Currently we focus on the REST catalog implementation but the above
> Catalog API proposal could work for other catalog types too. The internal
> implementation could be different, though. Initially they could throw a
> NotImplementedException.
> > >
> > >
> > > I hope this makes sense and I haven't missed any details or previous
> comments.
> > >
> > >
> > > Regards,
> > > Gabor
> > >
> > >
> > >
> > >
> > > On Tue, Nov 19, 2024 at 5:17 AM Taeyun Kim <
> taeyun....@innowireless.com> wrote:
> > >
> > > Hi,
> > >
> > > Here are my thoughts:
> > >
> > > - HTTP Layer: To my knowledge, there isn’t a separate "HTTP layer" in
> this context, so concerns about control over caching shouldn’t be an issue.
> The header approach I mentioned simply involves handling additional headers
> when using HTTP client libraries to interact with the REST API.
> > >
> > > - Programming Language: For reference, I don’t use Java - I use Rust
> and C++. Personally, I hope Iceberg’s specifications avoid including
> Java-specific features and that the cross-language compatible REST catalog
> becomes the primary catalog for Iceberg.
> > >
> > > - API Perspective: Given the above, I may not be in the best position
> to comment on Java APIs. However, regarding Gabor’s proposed API (Table
> loadTable(Table existingTable)), I believe it would be difficult for an
> ETag-based REST catalog to support this API since it cannot provide an
> ETag. Instead, I’d like to suggest an alternative API:
> > > Option<Table, tag> loadTableIfNoneMatch(TableIdentifier, Option<tag>)
> > > Initially, the client would provide None as the tag. If the tag is not
> None and matches the latest table tag, the API returns None (= not
> updated). If the tag is None or does not match the latest table tag, the
> API returns a new (Table, tag) pair.
> > >
> > > Thank you.
> > >
> > >
> > > -----Original Message-----
> > > From: "Zoltán Borók-Nagy" <borokna...@cloudera.com.invalid>
> > > To: <dev@iceberg.apache.org>;
> > > Cc:
> > > Sent: 2024-11-19 (화) 03:16:05 (UTC+09:00)
> > > Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the
> latest
> > >
> > >
> > > Hey Everyone,
> > >
> > >
> > > Thanks Gábor, I think the proposed interface would be very useful to
> any engine that employs caching, e.g. Impala.
> > > And it is pretty neat that it is catalog-agnostic, i.e. we just give
> all the information we have about the table and let the catalog
> implementation efficiently reload it.
> > >
> > >
> > > I might have a nitpick suggestion about the name to clearly express
> the intent: loadTable -> reloadTable (or, refreshTable)
> > >
> > >
> > > Cheers,
> > > Zoltan
> > >
> > >
> > >
> > >
> > > On Mon, Nov 18, 2024 at 5:17 PM Gabor Kaszab
> > > <gaborkas...@cloudera.com.invalid>
> wrote:
> > >
> > > Hi Iceberg Community,
> > >
> > >
> > > This is a great conversation so far, and thanks everyone for the
> valuable inputs!
> > > I'd like to articulate 2 things that we have to keep in mind with the
> design:
> > >
> > >
> > > 1: There are 2 interfaces here that we should consider:
> > > What I mean by this is that so far we have been talking about the REST
> spec, more narrowly the HTTP communication between Iceberg's REST client
> and the REST server. I think the proposed solution with the ETag absolutely
> makes sense within this context.
> > > However, the usual way of a client interacting with an Iceberg catalog
> (including REST) is the Catalog API in the library. This API offers a
> loadTable(TableIdentifier) function that returns a Table object. With the
> above HTTP-based solution in mind I don't think we could give any
> meaningful results if the HTTP layer finds that the table hasn't changed. I
> argued already against pushing the caching responsibilities from the
> clients into the HTTP layer (mostly because of losing the control over the
> cache, and also observability won't be straightforward) so let's assume for
> now that we won't do caching in the HTTP layer, only execute the loadTable
> calls to the REST catalog by setting the ETag. In case we get a 304 we
> won't be able to construct a Table object to answer the
> Catalog.loadTable(TableIdentifier) call. We could return null or throw an
> exception but I don't find any of them appropriate.
> > >
> > >
> > > 2: There are catalog types other than REST
> > > I started this conversation focusing on the REST spec, but the more I
> think of this the more I feel that the same functionality should be offered
> for all the other catalog types too. Let's assume that we have an engine
> that caches table metadata and initially uses REST catalog. For such an
> engine the proposed solution would solve the problem of checking table
> freshnes and also reloading the table metadata. A simple code for that
> could be enough if we configured our HTTP client properly (just sketched a
> simple example):
> > >
> > >
> > > tableCache_.put(catalog_.loadTable(tableIdentifier));
> > >
> > >
> > > Also let's assume we solve the issue in 1) and we can answer such a
> call even if we get 304 from the server as the table is unchanged. So with
> this solution with the REST catalog we can be sure that the table is only
> loaded from the catalog if changed (or the age expired). But what if we
> configure another catalog, let's say HiveCatalog. The very same code for
> that catalog would trigger a table reload for every execution causing
> unexpected performance issues.
> > > I have to double check but I assume that this HTTP approach wouldn't
> be feasible for other catalog types unfortunately.
> > >
> > >
> > > I hope these arguments make sense :)
> > >
> > >
> > >
> > >
> > > As a partial solution this is what I have in mind:
> > > We can add another function into the catalog API for this purpose.
> Let's say something like this:
> > > Table loadTable(Table existingTable);
> > >
> > >
> > > What advantages I see with this:
> > > - This could solve issue 1) above. In case the table hasn't changed we
> can simply return 'existingTable' without using HTTP Cache.
> > > - The clients wouldn't need to explicitly call for isLatest() and such
> functions to check for freshness, and they wouldn't need to trigger table
> reloading for themselve. This API would be expected to cover this under the
> hood.
> > > - The current Catalog.loadTable(TableIdentifier) API wouldn't be
> enough for all the catalog types on it's own, but with this one each
> catalog implementations (e.g. HiveCatalog, REST catalog, etc.) then can
> implement their own way of doing freshness checks and table reloads. For
> REST we could follow the HTTP ETag approach, while for other catalogs we
> could follow other approaches.
> > >
> > >
> > > Regards,
> > > Gabor
> > >
> > >
> > > On Mon, Nov 18, 2024 at 8:48 AM Shani Elharrar
> <sh...@upsolver.com.invalid> wrote:
> > >
> > > You're totally right. Perhaps using a "Content-Location" header might
> be a better fit for that.
> > >
> > >
> > > Shani.
> > >
> > >
> > > On 18 Nov 2024, at 9:27, Taeyun Kim <taeyun....@innowireless.com>
> wrote:
> > >
> > >
> > >
> > > Hi,
> > >
> > >
> > > Here are my thoughts:
> > >
> > >
> > > - The value of ETag is (as far as I know) defined as an opaque string
> by the specification, meaning the client shouldn’t interpret or assign any
> significance to it, regardless of what the server specifies. It’s best to
> avoid the client giving any particular meaning to the ETag value.
> > > - One major advantage of the header approach compared to other methods
> is that if an update has occurred, the updated content can be immediately
> included in the response without requiring an additional request. This
> saves one request-response round-trip (although It’s also possible to
> define a separate endpoint with the same functionality).
> > > - Since the Iceberg REST catalog server is effectively a type of HTTP
> server, at least in theory, it may be expected to handle HTTP cache and
> validation-related processes. The header approach can be seen as leveraging
> this mechanism appropriately.
> > > - The header approach doesn’t have to be limited to the
> /v1/{prefix}/namespaces/{namespace}/tables/{table} endpoint. It could also
> be applied to all GET-based endpoints, though this might broaden the scope
> significantly.
> > >
> > >
> > > Thank you.
> > >
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: "Shani Elharrar" <sh...@upsolver.com.invalid>
> > > To: <dev@iceberg.apache.org>;
> > > Cc: <dev@iceberg.apache.org>;
> > > Sent: 2024-11-18 (월) 16:21:16 (UTC+09:00)
> > > Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the
> latest
> > >
> > >
> > > Using the metadata file name as ETag is nice way to go. In that case,
> adding HEAD method support to the loadTable endpoint will return the latest
> metadata pointer, which can be used to support "isLatest" without returning
> the body. It can be also leveraged in order to return the latest metadata
> location of the table.
> > >
> > >
> > > Shani.
> > >
> > >
> > > On 18 Nov 2024, at 8:52, Yufei Gu <flyrain...@gmail.com> wrote:
> > >
> > >
> > >
> > >
> > > Hi Taeyun,
> > >
> > > Thank you for the clear explanation.
> > >
> > > I agree that the ETag solution is more suitable. If we were going that
> way, I'd propose a customized version number as an ETag—for instance,
> leveraging the metadata.json file name as the identifier.
> > >
> > > To summarize, HTTP caching relies on headers (e.g., ETag or
> Last-Modified) to validate whether a version is up-to-date, whereas the
> alternative approach proposed above uses an additional parameter for
> verification. From my perspective, there isn’t a fundamental difference
> between the two, so I’m OK with either.
> > >
> > > A couple of points to note:
> > >
> > > Both approaches would require changes to the "loadTable" endpoint.
> > > A minor advantage of HTTP caching is that it integrates seamlessly
> with browsers, but since most clients of the Iceberg REST catalog aren’t
> browsers, this may not be a significant factor.
> > > I’d also recommend considering the requirement to retrieve multiple
> tables(e.g., all tables under a namespace, or a list of table names) from
> the catalog. This requires a new endpoint and may not work with HTTP
> caching.
> > >
> > > Let me know your thoughts or if there’s anything else to consider.
> > >
> > > Yufei
> > >
> > >
> > >
> > >
> > > On Sun, Nov 17, 2024 at 6:43 PM Taeyun Kim <
> taeyun....@innowireless.com> wrote:
> > > Hi,
> > >
> > > To Gabor:
> > > It doesn’t seem necessary to interpret HTTP caching literally in this
> context.
> > > Simply using the HTTP headers defined by HTTP caching to check the
> freshness of metadata should be sufficient.
> > > There’s no requirement for the client to duplicate or store cached
> HTTP responses.
> > >
> > > To Yufei:
> > > As I understand it, the client doesn’t send its own timestamp but
> instead uses the timestamp originally provided by the server in the
> Last-Modified header.
> > > Therefore, clock synchronization issues should not be a concern.
> > >
> > > Here’s the general flow of HTTP cache validation based on
> If-Modified-Since:
> > >
> > > - Client: initial request:
> > >
> > > GET (url) HTTP/1.1
> > >
> > > - Server response:
> > >
> > > HTTP/1.1 200 OK
> > > Last-Modified: (date1)
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (with response body)
> > >
> > > - Client: validation request:
> > >
> > > GET (url) HTTP/1.1
> > > If-Modified-Since: (date1)
> > >
> > > - Server response (if unchanged):
> > >
> > > HTTP/1.1 304 Not Modified
> > > Last-Modified: (date1)
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (without response body)
> > >
> > > - Server response (if updated):
> > >
> > > HTTP/1.1 200 OK
> > > Last-Modified: (date2)
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (with response body)
> > >
> > > However, using time-based freshness checks can present challenges,
> such as parsing time formats or synchronizing file update times across
> servers.
> > > To address these issues, HTTP cache validation based on ETag is also
> defined in the specification.
> > >
> > > Here’s the flow for ETag-based validation:
> > >
> > > - Client: initial request:
> > >
> > > GET (url) HTTP/1.1
> > >
> > > - Server response:
> > >
> > > HTTP/1.1 200 OK
> > > ETag: "(arbitrary string 1 generated by the server)"
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (with response body)
> > >
> > > - Client: validation request:
> > >
> > > GET (url) HTTP/1.1
> > > If-None-Match: "(arbitrary string 1 generated by the server)"
> > >
> > > - Server response (if unchanged):
> > >
> > > HTTP/1.1 304 Not Modified
> > > ETag: "(arbitrary string 1 generated by the server)"
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (without response body)
> > >
> > > - Server response (if updated):
> > >
> > > HTTP/1.1 200 OK
> > > ETag: "(arbitrary string 2 generated by the server)"
> > > Cache-Control: no-store, no-cache, max-age=0, must-revalidate,
> proxy-revalidate
> > > (with response body)
> > >
> > > The server can choose to use either If-Modified-Since or ETag for
> freshness validation.
> > > Alternatively, to simplify the implementation related to the Iceberg
> REST catalog, it might make sense to define only the more accurate
> ETag-based validation in the spec.
> > > For reference, RFC 9110 recommends specifying both ETag and
> Last-Modified. When both are provided, ETag takes precedence.
> > >
> > > Note on Cache-Control Headers:
> > > The Cache-Control values in the examples above are intended to ensure
> that the client validates freshness with the server on every request.
> Writing the header in this extended format is primarily to accommodate
> outdated HTTP/1.1 implementations. However, under the HTTP/1.1
> specification, the following is sufficient:
> > >
> > > Cache-Control: no-cache
> > >
> > > That’s all for now.
> > > Thank you.
> > >
> > >
> > > -----Original Message-----
> > > From: "Yufei Gu" <flyrain...@gmail.com>
> > > To: <dev@iceberg.apache.org>;
> > > Cc:
> > > Sent: 2024-11-16 (토) 02:51:05 (UTC+09:00)
> > > Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the
> latest
> > >
> > >
> > >
> > > How does HTTP caching handle desynchronized clocks between clients and
> the server?
> > >
> > > At t0, the client gets the latest table version.
> > > At t1, the server makes a new commit.
> > > At t2, the client sends a request with a timestamp t2, but due to
> desynchronization, it refers to t0.
> > >
> > > The server may reply with 304 Not Modified, causing the client to
> think its cache is up-to-date and miss the commit at t1.
> > >
> > >
> > >
> > > Yufei
> > >
> > >
> > >
> > >
> > > On Fri, Nov 15, 2024 at 6:37 AM Gabor Kaszab <gaborkas...@apache.org>
> wrote:
> > > Hi All,
> > >
> > >
> > > First of all it's great to see that there are others who could benefit
> from giving a solution to this problem. I appreciate all the comments and
> feedback so far.
> > > There were a number of different opinions, so let me start with
> summarizing the different topics that came up:
> > >
> > >
> > > New endpoint vs using an existing endpoint:
> > > Based on the answers (Fokko, Yufei) I had the impression that we
> should be careful when adding new REST endpoints, and we should examine the
> re-use of existing endpoints first. Let's do that then, and in case we
> don't find it feasible then we can still fall back to any of my initial
> proposals (isLatest() or metadataLocation()).
> > >
> > >
> > > Granularity of freshness checks:
> > > It was brought up (Dmitri) that we might not want to do the metadata
> freshness checks solely based on metadata location, but we should consider
> doing more granular freshness checks. I personally don't see much benefit
> of designing this solution like that, TBH, but seeing some use-cases could
> help us understand the motivation here.
> > > Let me share my opinion on some of the arguments:
> > >
> > >
> > > "A change in metadata location does not necessarily mean a change in
> metadata content"
> > >
> > >
> > > AFAIK whenever Iceberg creates a new metadata file there is some
> change in the metadata itself. There might not be a new snapshot, though in
> the cases of e.g. a schema/partition evolution. But even in these cases
> triggering a table reload could make sense to me (e.g. answering SHOW
> CREATE TABLE and similar queries). Additionally, I'd assume the number of
> metadata location changes that don't create a new snapshot is too
> negligible to optimize for.
> > > Dmitri, let me know if I misunderstood something.
> > >
> > >
> > > "it may still be beneficial to permit the client to ask for changes to
> specific areas of metadata"
> > >
> > > This seems like a use-case that the partial metadata loading proposal
> could solve. To identify the need to load a specific part of the metadata
> with partial metadata loading seems an overkill to design with my proposal,
> if this is what you have in mind. Also I found that the partial metadata
> loading proposal faces serious headwinds, so I wouldn't rely on it at the
> moment.
> > >
> > >
> > > Re-using tableExists
> > > I think there is a consensus here that tableExists returning a
> metadata location could work but seems more like a workaround and could be
> misleading for the users.
> > >
> > > Partial metadata loading could solve this:
> > > (Yufei) I agree, it would be perfect for my use-case and I'm following
> the discussion on the proposal. However, for me it seems, as I wrote above,
> that the proposal faces serious headwinds now and I honestly wouldn't
> expect a solution in the short term. But solving the freshness problems is
> a more urgent thing to solve, not just for myself and Impala but apparently
> to many other stakeholders in the community according to the interest on
> this thread.
> > > Hence, I propose to come up with a separate solution for freshness
> checks, and we can still move to using partial metadata loading once that's
> out.
> > >
> > >
> > > Use HTTPCache and If-Modified-Since with loadTable
> > > This solution seems to do the trick for us. Let me do some research
> myself to see if there are any difficulties implementing this. Currently, I
> have more questions than answers wrt this approach :)
> > > - The initial problem is to answer freshness questions for the cached
> tables on the client side. If we introduce HttpCaching wouldn't we
> introduce the same problem but on a different level of representation. We'd
> then need to decide the freshness/staleness of the cached data in the HTTP
> layer.
> > > - If we cache the HTTP responses for a loadTable then we essentially
> cache the content of the metadata.jsons including the snapshot and metadata
> log and everything, plus the snapshot list (and I think the manifests for
> the latest snapshot). I believe that the size of this can easily reach the
> low megabytes range in memory, so in total keeping them in the HTTP Cache
> for all the tables we have queried can easily mean that we keep a couple of
> GBs in memory just for this purpose.
> > > For engines that already cache table metadata wouldn't this mean that
> we will cache some parts of the metadata redundantly?
> > > - How would we decide what is the max-age of a cached table metadata
> in the HTTP Cache? Would it be configurable so that each engine could use
> whatever it prefers?
> > >
> > >
> > > Sorry if any of the questions doesn't make sense, I just want to make
> sure I understand all the aspects of this approach.
> > >
> > >
> > > An additional topic I have in mind:
> > > REST catalog vs other catalogs:
> > > Now we are focusing our discussion on the REST spec, but I think it
> would be beneficial to extend our focus and cover other catalog
> implementations too. I don't think that this problem of data freshness is
> specific to REST catalog, it could affect any table in any other catalog
> too.
> > >
> > >
> > > I'll continue my investigation wrt the proposals, I just wanted to
> flush out and sum up what we have now before the weekend.
> > >
> > >
> > > Regards,
> > > Gabor
> > >
> > >
> > >
> > >
> > > On Fri, Nov 15, 2024 at 10:16 AM Jean-Baptiste Onofré <j...@nanthrax.net>
> wrote:
> > > Hi,
> > >
> > > I like the idea and it makes sense. As soon as it's clearly stated in
> > > the spec (using If-Modified-Since header and 304 status code), it
> > > looks good to me.
> > >
> > > Thanks !
> > > Regards
> > > JB
> > >
> > > On Fri, Nov 15, 2024 at 1:58 AM Taeyun Kim <
> taeyun....@innowireless.com> wrote:
> > > >
> > > > Hi,
> > > >
> > > > (Apologies if this email is a duplicate. This is my third attempt.)
> > > >
> > > > I also need a way to ensure that my table data is up-to-date. For
> now, I’m handling this by setting an expiration period after which I fetch
> the data again, regardless of its freshness.
> > > >
> > > > Here are my thoughts on the current suggestions. Please correct me
> if I've misunderstood any of the points.
> > > >
> > > > - isLatest(): This function could be inefficient since it would
> require an additional round-trip to fetch the metadata if it’s not
> up-to-date. This would result in two round-trips overall, which seems
> suboptimal.
> > > > - metadataLocation(): This has a similar issue as isLatest(). BTW,
> according to the REST catalog API documentation for LoadTableResult schema,
> it states, "Clients can check whether metadata has changed by comparing
> metadata locations after the table has been created." (
> https://github.com/apache/iceberg/blob/3659ded18d50206576985339bd55cd82f5e200cc/open-api/rest-catalog-open-api.yaml#L3175)
> This suggests that if the metadata location has changed, the metadata can
> be considered updated.
> > > > - tableExists(): Based on the name, this function seems to serve a
> different purpose.
> > > >
> > > > Here is my suggestion:
> > > >
> > > > Since HTTP has built-in caching features (
> https://developer.mozilla.org/en-US/docs/Web/HTTP/Caching), and REST
> catalogs operate over HTTP, it seems natural to leverage HTTP caching
> mechanisms. For example, HTTP includes the If-Modified-Since header and the
> 304 Not Modified status code. Using this approach, we could achieve data
> freshness with a single round-trip, fetching updated data only if there are
> modifications.
> > > >
> > > > What do you think about defining the spec in this direction?
> > > >
> > > > Thank you.
> > > >
> > > >
> > > >
> > > >
> > > > -----Original Message-----
> > > > From: "Yufei Gu" <flyrain...@gmail.com>
> > > > To: <dev@iceberg.apache.org>;
> > > > Cc:
> > > > Sent: 2024-11-13 (수) 03:43:24 (UTC+09:00)
> > > > Subject: Re: [DISCUSS] REST: Way to query if metadata pointer is the
> latest
> > > >
> > > >
> > > >
> > > > Hi Gamber,
> > > >
> > > > Thanks for the proposal! Impala isn’t unique in needing this—I've
> seen similar requirements from other engines.
> > > >
> > > > As others pointed out, using the “tableExists” endpoint seems like a
> workaround. I don't consider it a permanent way forward. We could address
> this by either modifying the current load table endpoint or introducing a
> new one, but ideally, we should avoid adding endpoints for every specific
> need. With that, partial metadata loading seems like a strong approach
> here, we will need certain agreement though. I'd suggest the community
> consider the use cases seriously. We need a way forward.
> > > >
> > > > I’m also not too concerned about using metadata file paths to verify
> the latest table version; clients can simply extract metadata filenames,
> which include the UUID.
> > > >
> > > > Yufei
> > > >
> > > >
> > > >
> > > >
> > > > On Tue, Nov 12, 2024 at 7:46 AM Jean-Baptiste Onofré <
> j...@nanthrax.net> wrote:
> > > >
> > > > Hi Fokko
> > > >
> > > > I like the idea, but I think it's more a workaround and could be
> > > > confusing for users :)
> > > >
> > > > Regards
> > > > JB
> > > >
> > > > On Tue, Nov 12, 2024 at 2:53 PM Fokko Driesprong <fo...@apache.org>
> wrote:
> > > > >
> > > > > Hey Gabor,
> > > > >
> > > > > Thanks for raising this. While reading this, my first thought is
> to leverage the `tableExists` operation:
> > > > >
> https://github.com/apache/iceberg/blob/e3f39972863f891481ad9f5a559ffef093976bd7/open-api/rest-catalog-open-api.yaml#L1129-L1160
> > > > >
> > > > > This doesn't return anything today, but we could return a payload
> to the latest metadata.json.
> > > > >
> > > > > Looking forward to what others think.
> > > > >
> > > > > Kind regards,
> > > > > Fokko
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Op di 12 nov 2024 om 14:33 schreef Shani Elharrar
> <sh...@upsolver.com.invalid>:
> > > > >>
> > > > >> I recommend option (b), provided there is no partial metadata
> loading. We implemented option (b) internally to facilitate partial
> metadata loading, as we have tables with hundreds of thousands of
> snapshots. This results in metadata that occupies approximately 500 MB in
> memory (excluding the JsonNodes), which is a significant load for some of
> our services.
> > > > >>
> > > > >> Shani.
> > > > >>
> > > > >> On 12 Nov 2024, at 14:12, Gabor Kaszab <gaborkas...@apache.org>
> wrote:
> > > > >>
> > > > >> Hey Iceberg Community,
> > > > >>
> > > > >> Background:
> > > > >> Impala is designed in a way to cache the Iceberg table metadata
> (BaseTable objects in practice) for faster access. Currently, Impala is
> tightly coupled with HMS and in turn with the HiveCatalog, and in order to
> keep the cached table objects up-to-date there is a notification mechanism
> driven by HMS to notify Impala about any changes in the table metadata.
> > > > >> The Impala community is actively looking for ways to decouple HMS
> from Impala and provide a way to use Impala without the need for HMS, and
> get the Iceberg table metadata from other catalog Implementations mainly
> focusing now on REST catalogs.
> > > > >>
> > > > >> Problem to solve:
> > > > >> We identified a particular missing functionality in the current
> REST spec: For engines that cache table metadata currently there is no way
> to check if that table metadata is up-to-date or not, and whether the
> engine should reload the metadata for that table or not without getting a
> whole table object from the catalog. For this I think the REST catalog (but
> in fact I think this could apply to any other catalogs) should be able to
> answer a question like:
> > > > >> "Hi Catalog, I have this version of this table, is it up-to-date?"
> > > > >>
> > > > >> Proposal:
> > > > >> I've been following the discussion about partial metadata loading
> that could be also used to answer the above question, but I have the
> impression now that the conversation stopped making any progress.
> > > > >> So instead of waiting for partial metadata loading I propose to
> have an addition to the REST spec now to answer the question I raised above:
> > > > >>
> > > > >> a) boolean isLatest(TableIdentifier ident, String
> metadataLocation);
> > > > >> b) String metadataLocation(TableIdentifier ident);
> > > > >>
> > > > >> Any of the above 2 approaches could help engines to decide if
> they have to invalidate/reload particular table metadata in the cache. I
> personally would go for option a) but would be open to hear other opinions.
> > > > >>
> > > > >> I'd like to know if the community could support me extending the
> REST spec with any of the 2 options.
> > > > >>
> > > > >> Regards,
> > > > >> Gabor
> > > > >>
> > > > >>
> > > 
>
