Originally I put the first two digits of the object SHA-1 into the
start of a row key to try and spread the load of objects around a DHT
service. Unfortunately this tends to not work as well as I had hoped.
Servers reading a repository need to contact every node in a DHT
cluster if the cluster tries to evenly distribute the object rows.
This is a lot of connections, especially if the cluster has many
backend storage servers. If the library has an open connection
limit (possibly due to JVM file descriptor limitations) it may need
to open and close a lot of connections to access a repository,
rather than being able to reuse the same connection to a handful
of backend servers. This results in a lot of connection thrashing
for some DHT type databases, and is inefficient.
Some DHTs are able to operate even if part of the database space
is currently unavailable. For example, a DHT service might assign
some section of the key space to a node, and then fail that section
over to another node when the primary is noticed as being offline.
During that failover period that section of the key space is not
available, but other sections hosted by other backends are still
ready for service. Spreading keys all over the cluster makes it
likely that any single backend being temporarily down means the
entire cluster is down, rather than only some.
This is a massive schema change, but it should improve relability
and performance for any DHT system.
Change-Id: I6b65bfb4c14b6f7bd323c2bd0638b49d429245be
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
jgit.storage.dht is a storage provider implementation for JGit that
permits storing the Git repository in a distributed hashtable, NoSQL
system, or other database. The actual underlying storage system is
undefined, and can be plugged in by implementing 7 small interfaces:
* Database
* RepositoryIndexTable
* RepositoryTable
* RefTable
* ChunkTable
* ObjectIndexTable
* WriteBuffer
The storage provider interface tries to assume very little about the
underlying storage system, and requires only three key features:
* key -> value lookup (a hashtable is suitable)
* atomic updates on single rows
* asynchronous operations (Java's ExecutorService is easy to use)
Most NoSQL database products offer all 3 of these features in their
clients, and so does any decent network based cache system like the
open source memcache product. Relying only on key equality for data
retrevial makes it simple for the storage engine to distribute across
multiple machines. Traditional SQL systems could also be used with a
JDBC based spi implementation.
Before submitting this change I have implemented six storage systems
for the spi layer:
* Apache HBase[1]
* Apache Cassandra[2]
* Google Bigtable[3]
* an in-memory implementation for unit testing
* a JDBC implementation for SQL
* a generic cache provider that can ride on top of memcache
All six systems came in with an spi layer around 1000 lines of code to
implement the above 7 interfaces. This is a huge reduction in size
compared to prior attempts to implement a new JGit storage layer. As
this package shows, a complete JGit storage implementation is more
than 17,000 lines of fairly complex code.
A simple cache is provided in storage.dht.spi.cache. Implementers can
use CacheDatabase to wrap any other type of Database and perform fast
reads against a network based cache service, such as the open source
memcached[4]. An implementation of CacheService must be provided to
glue this spi onto the network cache.
[1] https://github.com/spearce/jgit_hbase
[2] https://github.com/spearce/jgit_cassandra
[3] http://labs.google.com/papers/bigtable.html
[4] http://memcached.org/
Change-Id: I0aa4072781f5ccc019ca421c036adff2c40c4295
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Shawn O. Pearce
Matthias Sohn