On Tue, Jul 30, 2013 at 9:38 PM, James Taylor <jtay...@salesforce.com>wrote:
> So row key order won't match the natural sort order? > With the "Legacy" types that are based on Bytes, you get whatever you get. Strings and pass-through byte[]'s work like normal; positive integers will work, but not negatives; &c. HBase would not ship out-of-the-box with general purpose order-preserving types, but you'd have the API and be able to implement your own. On Tue, Jul 30, 2013 at 9:11 PM, Nick Dimiduk <ndimi...@gmail.com> wrote: > > > On Tue, Jul 30, 2013 at 8:20 PM, James Taylor <jtay...@salesforce.com > > >wrote: > > > > > What's the functionality that we'll lose without the order-preserving > > part > > > being included? > > > > > > > Well, order preservation ;) Lacking 8201, we'd get all the existing Bytes > > goodness but wrapped up in 8693's API. The framework is laid for other > > HBase components, user applications, and downstream projects to start > > building on it immediately. I'd like to start pluming it into some new > > Filters, a couple of the MapReduce tools, Hive's interop layer, maybe > even > > replace PDataType with DataType if you're open to such a patch. The > > OrderedBytes stuff will require applications to have a plan for data > > migration when they decide to transition over to it, but they can get a > > head-start on the boiler-plate code. > > > > On Tue, Jul 30, 2013 at 5:39 PM, Nick Dimiduk <ndimi...@gmail.com> > wrote: > > > > > > > On Tue, Jul 30, 2013 at 3:28 PM, Enis Söztutar <enis....@gmail.com> > > > wrote: > > > > > > > > > Let me elaborate. There are at least new RPC PB, PB structures in > > > HFiles > > > > / > > > > > hlogs, and zk, table locks, bucket cache, online merge, stochastic > > LB, > > > > > hbase on windows, *new data types*, AM changes, favorite node > > > assignment, > > > > > dist log replay, and tons of MTTR changes that are not run in > > > production > > > > so > > > > > far. > > > > > > > > > > > > Re: new data types, allow me to throw my hat into the ring as well. I > > > spoke > > > > with Stack this morning about a plan that I hope will allow the new > > data > > > > type API to squeeze in without requiring the order-preserving > encoding. > > > I'd > > > > like to get this API out with 0.96 so that users interested in > adopting > > > > this feature can start migrating their applications sooner than > later. > > > > Jealously, I want to start work on the Hive/HBase plumbing using this > > API > > > > and the Hive guys are only willing to accept code that is built > > against a > > > > labeled HBase release. I'm hoping other projects (*cough* Phoenix, > > > Impala, > > > > Kiji *cough*) are keen to follow suit. It also allows for the feature > > to > > > > fan out internally -- Filters, Coprocessors, ImportTsv, &c. Using > this > > > > approach, DataType and Legacy* friends come in for 0.96.0 and > > > OrderedBytes > > > > can join the party in a 0.96.x when it's ready for adoption. > > > > > > > > Thanks, > > > > Nick > > > > > > > > > >
