I just want to clarify the AWS dependency for queueing: Usergrid 2, at this time, depends on AWS for distributed queueing to allow for complete, fault tolerant multi-region indexing with Elasticsearch. There is an in-memory queueing service that works for single region deployments. You can also technically deploy the Usergrid 2 application itself in a multi-region fashion using the in-memory queuing for indexing. This deployment would consist of having a single region Elasticsearch cluster and each region's tomcat nodes pointing to the single ES cluster.
thanks! On Fri, Jan 15, 2016 at 11:15 AM, Dave <[email protected]> wrote: > Since early 2014, the Usergrid team has been working on a version of > Usergrid known informally as Usergrid 2. The effort was originally started > as a refactoring to clean up the codebase and make it easier to maintain, > but the work evolved into a new persistence engine for Usergrid. > > > The new Usergrid 2 engine is made up of a set of modules with clearly > defined interfaces. It uses Multi-version Concurrency Control instead of > distributed locks. It uses RxJava for concurrency. And the new engine uses > ElasticSearch to implement indexing and query. Usergrid 2 provides the > same REST API as Usergrid 1, but inside it is powered by this new engine. > > > Usergrid 2 was never released, but some have tested it, put it into > production, fixed lots of problems and hit some nice performance numbers, > for example: How Usergrid scaled to over 10,000 requests per second: > > > > http://apigee.com/about/blog/technology/how-usergrid-scaled-over-10000-requests-second > > > > The latest version of Usergrid 2 is informally known as Usergrid 2.1 and > you can find it in the Git “release” branch. Sidebar: state of Usergrid > branches: > > > - release: the well-tested “Usergrid 2.1” code base > - master: this is our development branch (with Jersey 2 and other > changes), currently called 2.1.1 > - 1.x: this is the 1.x code base, latest release was 1.0.2 > > I propose that we make a release from the release branch and call it Apache > Usergrid 2.1.0. I’d also like to make a convenience binary release with > the stack (i.e. ROOT.war), portal and an updated installation guide. > > > There are some drawbacks to Usergrid 2: it is more difficult to run in > production due to the addition of ElasticSearch, it depends on AWS SQS for > queueing and there are no tools for migrating from Usergrid 1.x — but > despite that, in my view, Usergrid 2.1.0 is ready for release and what we > want people to be using. > > > I'm not calling a vote, that will come later, but I would like to know your > thoughts and concerns about Usergrid 2 and this proposal. > > > Thanks, > > Dave >
