I am interested about your solution. Here is the detailed
architecture of my distributed vertical crawler.
http://www.flickr.com/photos/114261973@N07/
hope you guys can give me some advices.
1. goal
I want to implement a distributed vertical(topical) crawler. it
will only store webpages of a certan topic. I will have a classifier
to do this.
I estimated the amount of webpages that need be store is about
tens of millions(maybe hundreds of millions as time goes).
for vertical crawler, it should crawl the pages most likely
related to my target topics. So I need a frontier that can dispatch
task by priorities.
for now, the priority is simple but we hope it can deal with
complicated priority algorithms.
1.1 host priority
we should crawl many hosts rather than only one single host
at the same time. initally, each hosts should be equally crawled. but
after time, we can calculate the priority of host dynamically
e.g. we can control the speed of a certain host by it's
crawl history(some site will ban our crawler if we use too many
concurrent thread to it). or we can adjust the priority of a host by
whether it
is relevant to our topic(we can calculate the relevance of
crawled page).
1.2 enqueue time
first enqueued webpages should get higher priority
1.3 depth
webpages with small depth will get higher priority(something
like BFS traverse)
1.4 other page priorities
e.g. page rank, list page/detail page ...
2. archeitecture
see picture: http://www.flickr.com/photos/114261973@N07/
2.1 Seed Discover
use google or other website to find some seed urls
2.2 Url DB
a distributed DB to store all metadata about urls(that's the
most hbase related)
2.3 Task Scheduler
as described before, the task scheduler select top N priority
webpages and dispatch them to fetcher clusters
2.4 Message Queues
we use ActiveMQ to decouple different modules and also load balance
2.5 Fetchers
Download webpages
2.6 WebPageDB
store webpages crawled and extracted metadata(such as
title,content, pub_time, author, etc ....) of this webpage. we
consider using hbase too.
2.7 Extractors
Using classifier to judge whether this page is related to
our topics and extracting metadata from it and store them to WebPageDB
3. main challenges
3.1 Url DB
as described before, this store(maybe hbase) should support
sophisticated pirority algorithms. and also we use it to avoid
crawling a webpage more than once.
3.2 task scheduler
how to achieve our goal
4. current solution
4.1 use hbase(maybe together with phoenix) to store urls(we now
have not done the schema design, hoping get some advice here)
4.2 scheduler algorithm
int batchSize=10000;
//dispatch batchSize tasks to different hosts by host priorities;
Map<String,Integer> hostCount=...
//select top priority urls from each host
List<String> toBeCrawledUrls=new ArrayList<String>(batchSize);
for(Entry<String,Integer> entry:hostCount.entrySet()){
//select top priority N urls from a given host
List<String>
urls=selectTopNUrlsFromHost(entry.getKey(), entry.getValue());
toBeCrawledUrls.addAll(urls);
}
//dispatch this urls to message queue
//monitor the message queue status
//if the queue is all(or 3/4) consumed, goto top and
dispatch another batch urls
5. using map-reduce or hbase?
we discussed the possible usage of map-reduce or only hbase
if the scheduling algorithm is very complicated and should
consider many things, maybe we should use map-reduce
But for now, our algorithm is simple and using hbase
coprocesser(or phoenix) can be thought of a simple online map-reduce
we can use coprocesser to implement simple aggregating function
or using phoenix sql like select count where group by having....
On Fri, Jan 3, 2014 at 8:19 PM, Jean-Marc Spaggiari
<jean-m...@spaggiari.org> wrote:
> Interesting. This is exactly what I'm doing ;)
>
> I'm using 3 tables to achieve this.
>
> One table with the URL already crawled (80 millions), one URL with the URL
> to crawle (2 billions) and one URL with the URLs been processed. I'm not
> running any SQL requests against my dataset but I have MR jobs doing many
> different things. I have many other tables to help with the work on the
> URLs.
>
> I'm "salting" the keys using the URL hash so I can find them back very
> quickly. There can be some collisions so I store also the URL itself on the
> key. So very small scans returning 1 or something 2 rows allow me to
> quickly find a row knowing the URL.
>
> I also have secondary index tables to store the CRCs of the pages to
> identify duplicate pages based on this value.
>
> And so on ;) Working on that for 2 years now. I might have been able to use
> Nuthc and others, but my goal was to learn and do that with a distributed
> client on a single dataset...
>
> Enjoy.
>
> JM
>
>
> 2014/1/3 James Taylor <jtay...@salesforce.com>
>
>> Sure, no problem. One addition: depending on the cardinality of your
>> priority column, you may want to salt your table to prevent hotspotting,
>> since you'll have a monotonically increasing date in the key. To do that,
>> just add " SALT_BUCKETS=<n>" on to your query, where <n> is the number of
>> machines in your cluster. You can read more about salting here:
>> http://phoenix.incubator.apache.org/salted.html
>>
>>
>> On Thu, Jan 2, 2014 at 11:36 PM, Li Li <fancye...@gmail.com> wrote:
>>
>> > thank you. it's great.
>> >
>> > On Fri, Jan 3, 2014 at 3:15 PM, James Taylor <jtay...@salesforce.com>
>> > wrote:
>> > > Hi LiLi,
>> > > Have a look at Phoenix (http://phoenix.incubator.apache.org/). It's a
>> > SQL
>> > > skin on top of HBase. You can model your schema and issue your queries
>> > just
>> > > like you would with MySQL. Something like this:
>> > >
>> > > // Create table that optimizes for your most common query
>> > > // (i.e. the PRIMARY KEY constraint should be ordered as you'd want
>> your
>> > > rows ordered)
>> > > CREATE TABLE url_db (
>> > > status TINYINT,
>> > > priority INTEGER NOT NULL,
>> > > added_time DATE,
>> > > url VARCHAR NOT NULL
>> > > CONSTRAINT pk PRIMARY KEY (status, priority, added_time, url));
>> > >
>> > > int lastStatus = 0;
>> > > int lastPriority = 0;
>> > > Date lastAddedTime = new Date(0);
>> > > String lastUrl = "";
>> > >
>> > > while (true) {
>> > > // Use row value constructor to page through results in batches of
>> > 1000
>> > > String query = "
>> > > SELECT * FROM url_db
>> > > WHERE status=0 AND (status, priority, added_time, url) > (?, ?,
>> > ?,
>> > > ?)
>> > > ORDER BY status, priority, added_time, url
>> > > LIMIT 1000"
>> > > PreparedStatement stmt = connection.prepareStatement(query);
>> > >
>> > > // Bind parameters
>> > > stmt.setInt(1, lastStatus);
>> > > stmt.setInt(2, lastPriority);
>> > > stmt.setDate(3, lastAddedTime);
>> > > stmt.setString(4, lastUrl);
>> > > ResultSet resultSet = stmt.executeQuery();
>> > >
>> > > while (resultSet.next()) {
>> > > // Remember last row processed so that you can start after that
>> > for
>> > > next batch
>> > > lastStatus = resultSet.getInt(1);
>> > > lastPriority = resultSet.getInt(2);
>> > > lastAddedTime = resultSet.getDate(3);
>> > > lastUrl = resultSet.getString(4);
>> > >
>> > > doSomethingWithUrls();
>> > >
>> > > UPSERT INTO url_db(status, priority, added_time, url)
>> > > VALUES (1, ?, CURRENT_DATE(), ?);
>> > >
>> > > }
>> > > }
>> > >
>> > > If you need to efficiently query on url, add a secondary index like
>> this:
>> > >
>> > > CREATE INDEX url_index ON url_db (url);
>> > >
>> > > Please let me know if you have questions.
>> > >
>> > > Thanks,
>> > > James
>> > >
>> > >
>> > >
>> > >
>> > > On Thu, Jan 2, 2014 at 10:22 PM, Li Li <fancye...@gmail.com> wrote:
>> > >
>> > >> thank you. But I can't use nutch. could you tell me how hbase is used
>> > >> in nutch? or hbase is only used to store webpage.
>> > >>
>> > >> On Fri, Jan 3, 2014 at 2:17 PM, Otis Gospodnetic
>> > >> <otis.gospodne...@gmail.com> wrote:
>> > >> > Hi,
>> > >> >
>> > >> > Have a look at http://nutch.apache.org . Version 2.x uses HBase
>> > under
>> > >> the
>> > >> > hood.
>> > >> >
>> > >> > Otis
>> > >> > --
>> > >> > Performance Monitoring * Log Analytics * Search Analytics
>> > >> > Solr & Elasticsearch Support * http://sematext.com/
>> > >> >
>> > >> >
>> > >> > On Fri, Jan 3, 2014 at 1:12 AM, Li Li <fancye...@gmail.com> wrote:
>> > >> >
>> > >> >> hi all,
>> > >> >> I want to use hbase to store all urls(crawled or not crawled).
>> > >> >> And each url will has a column named priority which represent the
>> > >> >> priority of the url. I want to get the top N urls order by
>> > priority(if
>> > >> >> priority is the same then url whose timestamp is ealier is
>> prefered).
>> > >> >> in using something like mysql, my client application may like:
>> > >> >> while true:
>> > >> >> select url from url_db order by priority,addedTime limit
>> > >> >> 1000 where status='not_crawled';
>> > >> >> do something with this urls;
>> > >> >> extract more urls and insert them into url_db;
>> > >> >> How should I design hbase schema for this application? Is
>> hbase
>> > >> >> suitable for me?
>> > >> >> I found in this article
>> > >> >>
>> > >>
>> >
>> http://blog.semantics3.com/how-we-built-our-almost-distributed-web-crawler/
>> > >> >> ,
>> > >> >> they use redis to store urls. I think hbase is originated from
>> > >> >> bigtable and google use bigtable to store webpage, so for huge
>> number
>> > >> >> of urls, I prefer distributed system like hbase.
>> > >> >>
>> > >>
>> >
>>
