HyperDex 1.2 Release Announcement: HyperDex becomes a Document Store

[Robert Escriva]

We are proud to announce HyperDex 1.2, the next generation NoSQL data
store that provides strong consistency, fault-tolerance, and
high-performance.

The key features added by this release are:

- Document storage: Store JSON objects in HyperDex and
query them without having to define a schema before hand.

- Dynamic indices: Add and remove indices on attributes and documents
even after loading data into HyperDex.

- Node.JS bindings; This release adds fully-supported Node.JS
bindings.

HyperDex runs on 64-bit Linux (Ubuntu, Debian, Fedora, Centos) and OS X.
Binary packages for Debian 7, Ubuntu 12.04-13.10, Fedora 18-19, and
CentOS 6 are available[1], as well as source tarballs for other Linux
platforms.

HyperDex provides bindings for C, C++, Python, Java, Ruby, Go, and Node.

Happy Hacking,
The HyperDex Team

[1] http://hyperdex.org/download

[Robert Escriva]

Hi Roja,

I'd be happy to elaborate on your questions:

Performance of documents:
Right now, we have no publishable benchmarks here, but
qualitatively, documents feel like most other data types and are not
noticeably different performance-wise. Keep in mind that this is
the first release with documents, so performance can only get better
from here.

Indices/objects and hyperspaces:
Hyperspace hashing efficiently maps objects to servers. The indices
we mention make queries efficient on individual servers, and do not
alter the way in which objects map to servers.

So indices and hyperspace hashing (or subspaces) go hand in hand.
Hyperspace hashing maps your query onto a subset of servers in the
system and indexing enables those servers to efficiently respond to
queries.

Documents in Warp:
HyperDex Warp supports the same operations on documents that
HyperDex itself does. You can fetch and update documents within a
transaction just as you would outside a transaction. This enables
you to safely change multiple documents atomically with all the
guarantees Warp upholds.

-Robert

On Tue, Apr 22, 2014 at 06:24:32AM -0700, Roja Buck wrote:
> Would it be possible to give an overview of:
>
> The performance characteristics of the document based objects?
> How the indexes / objects relate to the hyper space model?
> How Warp interacts with documents?
>
> Regards,
>
> Roja

> [1] [1]http://hyperdex.org/download
>
> --
> You received this message because you are subscribed to the Google Groups
> "hyperdex-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to [2]hyperdex-discu...@googlegroups.com.
> For more options, visit [3]https://groups.google.com/d/optout.
>
> References:
>
> [1] http://hyperdex.org/download
> [2] mailto:hyperdex-discu...@googlegroups.com
> [3] https://groups.google.com/d/optout