Re: [ZODB-Dev] Using zodb and blobs
On 04/15/2010 06:00 PM, Adam GROSZER wrote: Hello Christian, Wednesday, April 14, 2010, 8:30:50 AM, you wrote: CT I don't think the transfer rate is actually that interesting. For small CT but many transactions the seek time/spinning speed should have the CT limiting influence. CT I've run the attached script a couple of times on my notebook, here's CT the results: CT 0.11 909.090909091 CT 0.15 666.7 CT 0.2 500.0 CT 0.07 1428.57142857 CT 0.07 1428.57142857 CT 0.14 714.285714286 CT The initial runs are a bit lower as they were interfered with by other CT applications writing to the disk. CT It's a notebook w/ Intel P9600, Seagate 7.2k SATA drive, 4GB RAM, Ubuntu CT 10.04, linux 2.6.32, ext4 Something is wrong with the seek time. Trying it on an Intel G1 80G SSD goes hardly over a 1000. Well, sounds like a broken fsync then. :/ -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development smime.p7s Description: S/MIME Cryptographic Signature ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Hello Christian, Wednesday, April 14, 2010, 8:30:50 AM, you wrote: CT I don't think the transfer rate is actually that interesting. For small CT but many transactions the seek time/spinning speed should have the CT limiting influence. CT I've run the attached script a couple of times on my notebook, here's CT the results: CT 0.11 909.090909091 CT 0.15 666.7 CT 0.2 500.0 CT 0.07 1428.57142857 CT 0.07 1428.57142857 CT 0.14 714.285714286 CT The initial runs are a bit lower as they were interfered with by other CT applications writing to the disk. CT It's a notebook w/ Intel P9600, Seagate 7.2k SATA drive, 4GB RAM, Ubuntu CT 10.04, linux 2.6.32, ext4 Something is wrong with the seek time. Trying it on an Intel G1 80G SSD goes hardly over a 1000. NB, P7400 3GB RAM, ubuntu 9.10, ext4 0.14 714.285714286 0.14 714.285714286 0.1 1000.0 0.14 714.285714286 Just for fun, it's flying on a tmpfs 0.36 2777.7778 0.37 2702.7027027 0.37 2702.7027027 Vmware makes it fly also: with 1000 runs on a 4GHz E8400, ubuntu 9.10, ext4 within VMware 0.24 4166.6667 0.33 3030.3030303 0.26 3846.15384615 0.31 3225.80645161 0.29 3448.27586207 in the VM tmpfs does not boost a lot: 0.2 5000.0 0.2 5000.0 0.2 5000.0 0.19 5263.15789474 0.19 5263.15789474 -- Best regards, Adam GROSZERmailto:agros...@gmail.com -- Quote of the day: People seem to enjoy things more when they know a lot of other people have been left out of the pleasure. - Russell Baker ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
On Tue, Apr 13, 2010 at 8:42 PM, Nitro ni...@dr-code.org wrote: Am 14.04.2010, 04:08 Uhr, schrieb Laurence Rowe l...@lrowe.co.uk: Running your test script on my small amazon EC2 instance on linux takes between 0.0 and 0.04 seconds (I had to remove the divide by total to avoid a zero division error). 0.02 is 5000/s. I don't know how EC2 works in detail, but 5000 transactions per second sound impossible to write to disk. Even 500 are impossible if your disk doesn't have VERY fast access times. Unlike most other databases, ZODB records written to file storages are always appended, so there is no seeking involved. The only seeking involved in writes is that needed to read previous records, but if a test is simply writing the same object over and over, or updating a small corpus, the previous record is likely to be in disk cache. Of course, other things happening on the system will typically cause the disk heads to seek away from the end of the database file, but you're unlikely to see that in a simpler benchmark. Jim -- Jim Fulton ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
On 04/14/2010 03:30 AM, Nitro wrote: Am 14.04.2010, 04:39 Uhr, schrieb Tim Peterstim.pet...@gmail.com: [Nitro] ... I wonder if _commit is really *that* slow Six years ago I timed factor-of-100 speed differences due to using MS _commit() on WinXP at the time: https://mail.zope.org/pipermail/zodb-dev/2004-July/007720.html or if there's another (faster) function which can be called... No MS function that does the same thing. Seems like ZODB has a long history of discussions on this matter: http://www.mail-archive.com/zodb-dev@zope.org/msg01874.html There's even a proposal for improvement in that thread, also on the wiki: http://wiki.zope.org/ZODB/FsyncBehaviourSetting What I don't really get is why you should never use None on windows. As far as I can judge from the various transaction rates in the thread Tim mentioned, fsync is just a no-op on linux anyways (depending on the specific file system of course). I'm pretty sure it's not. IIRC fsync is defined by POSIX and absolutely requires the implementor to flush data physically to disk ensuring its persistency. If that doesn't hold true then all transactions are borked. I've seen virtualised environments like VMWare ESX lie about fsyncs from a virtual hardware perspective. A similar issue with breaking fsync was the discussion around ext4 on notebooks with a only actually flush discs every 10 seconds. In the end it really depends on what you need your data for. If I'd store the information that a customer paid me 50 EUR for something and I presented a screen to him that told him he'll receive some good for that then I'd rather stick with compliant transactions. I am almost tempted to do os.fsync = lambda fd: 0 and rely on yesterday's backup. 0.49 j/k. j/k? -- Christian Theune · c...@gocept.com gocept gmbh co. kg · forsterstraße 29 · 06112 halle (saale) · germany http://gocept.com · tel +49 345 1229889 0 · fax +49 345 1229889 1 Zope and Plone consulting and development ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
On 4/14/10 08:24 , Christian Theune wrote: I'm pretty sure it's not. IIRC fsync is defined by POSIX and absolutely requires the implementor to flush data physically to disk ensuring its persistency. If that doesn't hold true then all transactions are borked. That was the problem with fsync on Linux: it effectively flushed all pending filesystem work, not just that for your current filehandle. That was needed to satisfy ordering constraints for the filesystem. And even though the result might be a lie since disks or other bits of hardwire can lie to you. It is generally better to use fdatasync() instead of fsync(), but you could still end up waiting much longer than you would expect. Wichert. ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Am 14.04.2010, 09:24 Uhr, schrieb Christian Theune c...@gocept.com: What I don't really get is why you should never use None on windows. As far as I can judge from the various transaction rates in the thread Tim mentioned, fsync is just a no-op on linux anyways (depending on the specific file system of course). I'm pretty sure it's not. IIRC fsync is defined by POSIX and absolutely requires the implementor to flush data physically to disk ensuring its persistency. If that doesn't hold true then all transactions are borked. They are. See https://mail.zope.org/pipermail/zodb-dev/2004-July/007683.html . Important quote Linux will allow fsync() to return even if the data isn't actually on disk. Also you can do the math yourself. If you have 10 ms average seek time, then you can only do 100 fsync/s. If you get more, there's buffering of some sort. Be it in hardware or file system. Since you cannot rely on fsync to work here I see the fsync as an hopeful wish, but nothing you can rely on. Things you cannot rely on give a false sense of security even if they might actually help in some cases. That's why I trust Windows' _commit to do it's work much more I trust linux' fsync to do it's job. 50 transaction = 20ms per transaction which sounds reasonable. In the end it really depends on what you need your data for. If I'd store the information that a customer paid me 50 EUR for something and I presented a screen to him that told him he'll receive some good for that then I'd rather stick with compliant transactions. Yes, in my case it's nothing critical or related to money. If there's a hardware outage a day of work is lost at worst. In case of corruption (which can happen also without fsync as data within the file can just be garbled) you need a backup anyways. j/k? just kidding. I don't think the transfer rate is actually that interesting. For small but many transactions the seek time/spinning speed should have the limiting influence. Yes, seek time is important here. I didn't recall the seek times of my hard disk, but wanted to mention it's not a slow hard disk. -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
On Wed, Apr 14, 2010 at 11:52 AM, Nitro ni...@dr-code.org wrote: Yes, in my case it's nothing critical or related to money. If there's a hardware outage a day of work is lost at worst. In case of corruption (which can happen also without fsync as data within the file can just be garbled) you need a backup anyways. Usually you will only loose the last transaction and not a days of work. The Data.fs is an append-only file, with one transaction appended after another. If there's a garbled or incomplete write, you'll typically loose the last transaction. The ZODB is smart enough to detect broken transactions and skip them on restart. I have witnessed one ZEO installation myself, where the physical machine hosting the ZEO server restarted multiple times a day, over a period of months. Nobody noticed for a long time, as the application was accessible all the time and no data had been lost. Obviously this wasn't a very write-intense application. But it still showed me how stable the ZODB really is. Hanno ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Am 14.04.2010, 14:45 Uhr, schrieb Hanno Schlichting ha...@hannosch.eu: Usually you will only loose the last transaction and not a days of work. The Data.fs is an append-only file, with one transaction appended after another. If there's a garbled or incomplete write, you'll typically loose the last transaction. The ZODB is smart enough to detect broken transactions and skip them on restart. I have witnessed one ZEO installation myself, where the physical machine hosting the ZEO server restarted multiple times a day, over a period of months. Nobody noticed for a long time, as the application was accessible all the time and no data had been lost. Obviously this wasn't a very write-intense application. But it still showed me how stable the ZODB really is. Yes, I agree with your opinion in general. There's still a chance that broken transactions are written IIUC (see https://mail.zope.org/pipermail/zodb-dev/2004-July/007683.html ): Doesn't this mean that if the system suddenly crashes in the middle of os.fsync, the Data.fs on disk will contain an incomplete transaction, but the transaction status byte would claim that the transaction is complete. Wouldn't that be bad? If that happened, perhaps. The chance exists because fsync does not work as advertised on many systems. On the systems where it seems to work, the slowdown is massive. So I am doubting the usefulness of using fsync in current ZODB at all. As your observations seem to hint, it's probably very unlikely to encounter this problem in practice. And I doubt Tim finally got Jim to pay him for a 48 hour pull-the-plug session :-) That's why I am not going to dig further into this and am satisfied with the current reliability. -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Hello Tres, thanks for your detailed answers! Am 12.04.2010, 22:42 Uhr, schrieb Tres Seaver tsea...@palladion.com: Additionally I made some quick performance tests. I committed 1kb sized objects and I can do about 40 transaction/s if one object is changed per transaction. For 100kb objects it's also around 40 transactions/s. Only for object sizes bigger than that the raw I/O throughput seems to start to matter. 40 tps sounds low: are you pushing blob content over the wire somehow? No, that test was with a plain file storage. Just a plain Persistent object with a differently sized string and an integer attribute. I did something like 1) create object with attribute x (integer) and y (variably sized string) 2) for i in range(100): obj.x = i; transaction.commit() 3) Measure time taken for step 2 Still don't know the answers to these: - Does it make sense to use ZODB in this scenario? My data is not suited well for an RDBMS. YMMV. I still default to using ZODB for anything at all, unless the problem smells very strongly relational. Ok, the problem at hand certainly doesn't smell relational. It is more about storing lots of different data than querying it extensively. It's a mixture of digital asset management (the blobs are useful for this part) and projects which reference the assets. The projects are shared between the clients and will consist of a big tree with Persistent objects hooked up to it. - Are there more complications to blobs other than a slightly different backup procedure? You need to think about how the blob data is shared between ZEO clients (your appserver) and the ZEO storage server: opinions vary here, but I would prefer to have the blobs living in a writable shared filesystem, in order to avoid the necessity of fetching their data over ZEO on the individual clients which were not the one pushing the blob into the database. The zeo server and clients will be in different physical locations, so I'd probably have to employ some shared filesystem which can deal with that. Speaking of locations of server and clients, is it a problem - as in zeo will perform very badly under these circumstances as it was not designed for this - if they are not in the same location (typical latency 0-100ms)? - Are there any performance penalties by using very large invalidation queues (i.e. 300,000 objects) to reduce client cache verification time? At a minimum, RAM occupied by that queue might be better used elsewhere. I just don't use persistent caches, and tend to reboot appservers in rotation after the ZEO storage has been down for any significant period (almost never happens). In my case the clients might be down for a couple of days (typically 1 or 2 days) and they should not spend 30 mins in cache verification time each time they reconnect. So if these 300k objects take up 1k each, then they occupy 300 MB of ram which I am fine with. From what I've read it only seems to consume memory. Note that the ZEO storage server makes copies of that queue to avoid race conditions. Ok, I can see how copying and storing 300k objects is slow and can take up excessive amounts of memory. Thanks, -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
40 tps sounds low: are you pushing blob content over the wire somehow? I have seen the ZEO storage committing transactions at least an order of magnitude faster than that (e.g., when processing incoming newswire feeds). I would guess that there could have been some other latencies involved in your setup (e.g., that 0-100ms lag you mention below). See my attached test script. It outputs 45-55 transactions/s for 100 byte sized payload. Maybe there's a very fundamental flaw in the way the test is setup. Note that I am testing on a regular desktop machine (Windows 7, WoW64, 4GB RAM, 1TB hard disk capable of transfer rates 100MB/s). The zeo server and clients will be in different physical locations, so I'd probably have to employ some shared filesystem which can deal with that. Speaking of locations of server and clients, is it a problem - as in zeo will perform very badly under these circumstances as it was not designed for this - if they are not in the same location (typical latency 0-100ms)? That depends on the mix of reads and writes in your application. I have personnally witnessed a case where the clients stayed up and serving pages over a whole weekend in a clusterfsck where both the ZEO server and the monitoring infrastructure went belly up. This was for a large corporate intranet, in case that helps: the problem surfaced mid-morning on Monday when the employee in charge of updating the lunch menu for the week couldn't save the changes. Haha, I hope they solved this critical problem in time! In my case the clients might be down for a couple of days (typically 1 or 2 days) and they should not spend 30 mins in cache verification time each time they reconnect. So if these 300k objects take up 1k each, then they occupy 300 MB of ram which I am fine with. If the client is disconnected for any period of time, it is far more likely that just dumping the cache and starting over fresh will be a win. The 'invalidation_queue' is primarily to support clients which remain up while the storage server is down or unreachable. Yes, taking the verification time hit is my plan for now. However, dumping the whole client cache is something I'd like to avoid, since the app I am working on will not work over a corporate intranet and thus the bandwidth for transferring the blobs is limited (and so can take up considerable time). Maybe I am overestimating the whole client cache problem though. Thanks again for your valuable advice, -Matthias test.py Description: Binary data ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Running your test script on my small amazon EC2 instance on linux takes between 0.0 and 0.04 seconds (I had to remove the divide by total to avoid a zero division error). 0.02 is 5000/s. Laurence On 14 April 2010 00:25, Nitro ni...@dr-code.org wrote: 40 tps sounds low: are you pushing blob content over the wire somehow? I have seen the ZEO storage committing transactions at least an order of magnitude faster than that (e.g., when processing incoming newswire feeds). I would guess that there could have been some other latencies involved in your setup (e.g., that 0-100ms lag you mention below). See my attached test script. It outputs 45-55 transactions/s for 100 byte sized payload. Maybe there's a very fundamental flaw in the way the test is setup. Note that I am testing on a regular desktop machine (Windows 7, WoW64, 4GB RAM, 1TB hard disk capable of transfer rates 100MB/s). The zeo server and clients will be in different physical locations, so I'd probably have to employ some shared filesystem which can deal with that. Speaking of locations of server and clients, is it a problem - as in zeo will perform very badly under these circumstances as it was not designed for this - if they are not in the same location (typical latency 0-100ms)? That depends on the mix of reads and writes in your application. I have personnally witnessed a case where the clients stayed up and serving pages over a whole weekend in a clusterfsck where both the ZEO server and the monitoring infrastructure went belly up. This was for a large corporate intranet, in case that helps: the problem surfaced mid-morning on Monday when the employee in charge of updating the lunch menu for the week couldn't save the changes. Haha, I hope they solved this critical problem in time! In my case the clients might be down for a couple of days (typically 1 or 2 days) and they should not spend 30 mins in cache verification time each time they reconnect. So if these 300k objects take up 1k each, then they occupy 300 MB of ram which I am fine with. If the client is disconnected for any period of time, it is far more likely that just dumping the cache and starting over fresh will be a win. The 'invalidation_queue' is primarily to support clients which remain up while the storage server is down or unreachable. Yes, taking the verification time hit is my plan for now. However, dumping the whole client cache is something I'd like to avoid, since the app I am working on will not work over a corporate intranet and thus the bandwidth for transferring the blobs is limited (and so can take up considerable time). Maybe I am overestimating the whole client cache problem though. Thanks again for your valuable advice, -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - zodb-...@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
Am 14.04.2010, 04:08 Uhr, schrieb Laurence Rowe l...@lrowe.co.uk: Running your test script on my small amazon EC2 instance on linux takes between 0.0 and 0.04 seconds (I had to remove the divide by total to avoid a zero division error). 0.02 is 5000/s. Thanks for running the test. Intrigued by this extreme difference I've done a little run with cProfile, script attached. On my machine the 100 runs take ~2.65 seconds, at least 2.55 seconds are spent in the nt.fsync function. That's an alias for os.fsync on windows. According to the python docs it calls the _commit C function on windows. See http://msdn.microsoft.com/en-us/library/17618685(VS.80).aspx . I wonder if _commit is really *that* slow or if there's another (faster) function which can be called... -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
[Nitro] ... I wonder if _commit is really *that* slow Six years ago I timed factor-of-100 speed differences due to using MS _commit() on WinXP at the time: https://mail.zope.org/pipermail/zodb-dev/2004-July/007720.html or if there's another (faster) function which can be called... No MS function that does the same thing. ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
For the record, with import os os.fsync = lambda fd: 0 at the top of the test app I get ~3700 tx/s. -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Nitro wrote: After lots of googling and browsing the source I can answer some of my questions: - What's the difference between storing bigger objects as blobs and as plain large strings? Plain large strings cannot be streamed for instance. Products like Zope chop up their file uploads into 64kb chunks which are then stored as individual objects in the zodb. That was the strategy before blobs. ZODB versions since 3.8 support storage of BLOb data as files on the filesystem. - Can I stream in parts of a blob/large string without having to read all of it? I can get a file handle to a blob. Strings are always read as a whole. - Where can I find example code on zodb blobs? E.g. how do I save a blob, how do I read it back in? The ZODB/tests directory features a few blob doctests which provide all the necessary code to get started. Having this on zodb.org would be nice (especially since the doctests are already ReST-formatted). Additionally I made some quick performance tests. I committed 1kb sized objects and I can do about 40 transaction/s if one object is changed per transaction. For 100kb objects it's also around 40 transactions/s. Only for object sizes bigger than that the raw I/O throughput seems to start to matter. 40 tps sounds low: are you pushing blob content over the wire somehow? Still don't know the answers to these: - Does it make sense to use ZODB in this scenario? My data is not suited well for an RDBMS. YMMV. I still default to using ZODB for anything at all, unless the problem smells very strongly relational. - Are there more complications to blobs other than a slightly different backup procedure? You need to think about how the blob data is shared between ZEO clients (your appserver) and the ZEO storage server: opinions vary here, but I would prefer to have the blobs living in a writable shared filesystem, in order to avoid the necessity of fetching their data over ZEO on the individual clients which were not the one pushing the blob into the database. - Is it ok to use cross-database references? Or is this better avoided at all cost? I would normally avoid them out of habit. They seem to work, though. And new questions: - Does the _p_invalidate hooking as outlined at http://www.mail-archive.com/zodb-dev@zope.org/msg00637.html work reliably? Never tried it, nor felt the need. - Are there any performance penalties by using very large invalidation queues (i.e. 300,000 objects) to reduce client cache verification time? At a minimum, RAM occupied by that queue might be better used elsewhere. I just don't use persistent caches, and tend to reboot appservers in rotation after the ZEO storage has been down for any significant period (almost never happens). From what I've read it only seems to consume memory. Note that the ZEO storage server makes copies of that queue to avoid race conditions. Tres. - -- === Tres Seaver +1 540-429-0999 tsea...@palladion.com Palladion Software Excellence by Designhttp://palladion.com -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkvDd5sACgkQ+gerLs4ltQ5h4wCghqTQsNO/5XrqHMZyhi8Hz17d oRcAn1el86604KoTTWB8Bx5R13ZlvQB/ =momg -END PGP SIGNATURE- ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Using zodb and blobs
After lots of googling and browsing the source I can answer some of my questions: - What's the difference between storing bigger objects as blobs and as plain large strings? Plain large strings cannot be streamed for instance. Products like Zope chop up their file uploads into 64kb chunks which are then stored as individual objects in the zodb. - Can I stream in parts of a blob/large string without having to read all of it? I can get a file handle to a blob. Strings are always read as a whole. - Where can I find example code on zodb blobs? E.g. how do I save a blob, how do I read it back in? The ZODB/tests directory features a few blob doctests which provide all the necessary code to get started. Having this on zodb.org would be nice (especially since the doctests are already ReST-formatted). Additionally I made some quick performance tests. I committed 1kb sized objects and I can do about 40 transaction/s if one object is changed per transaction. For 100kb objects it's also around 40 transactions/s. Only for object sizes bigger than that the raw I/O throughput seems to start to matter. Still don't know the answers to these: - Does it make sense to use ZODB in this scenario? My data is not suited well for an RDBMS. - Are there more complications to blobs other than a slightly different backup procedure? - Is it ok to use cross-database references? Or is this better avoided at all cost? And new questions: - Does the _p_invalidate hooking as outlined at http://www.mail-archive.com/zodb-dev@zope.org/msg00637.html work reliably? - Are there any performance penalties by using very large invalidation queues (i.e. 300,000 objects) to reduce client cache verification time? From what I've read it only seems to consume memory. Thanks, -Matthias ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org https://mail.zope.org/mailman/listinfo/zodb-dev