A Fantastic New Ride for WebLogic: Exalogic Elastic Cloud

Wow, finally a breather from the frenzy of OpenWorld, Develop, and JavaOne, and a chance to share some thoughts on a really exciting addition to the application grid story: Oracle Exalogic Elastic Cloud. The latest in a series of offerings that leverage Oracle’s new hardware capabilities acquired from Sun, Exalogic follows Exadata as a combined hardware-software solution designed to continue revolutionizing the datacenter. Exadata changed the game for running your database, and Exalogic changes the game for running your business logic.

Exalogic_top_closed_0056

You could think of Exalogic as a “middleware machine”, but that doesn’t completely do it justice in that, at the end of the day, its job is not to run middleware per se but rather to run applications. The way it achieves that however is by running middleware (and system software) extremely well—in particular, WebLogic Server, Coherence, JRockit, HotSpot, Oracle Linux, and Solaris. Exalogic is a rack containing 30 compute nodes with a massive storage subsystem backplaned by an extreme-performance InfiniBand interconnect and running a highly tuned WebLogic stack. The compute nodes are Sun Fire X4170 M2s, each with two sockets of 6-core 2.93 GHz 64-bit Intel Xeon processors, 960 GB of solid-state disk, and surrounding hardware to make them scream. Additional storage includes 40 TB of SAS disks with 4 TB of read cache and 72 GB of write cache. A big part of the magic comes from InfiniBand, a 40 Gb/sec switched fabric with numerous advantages over typical interconnect approaches. We’ve tuned the WebLogic stack to take advantage of this by eliminating many buffer copies, transferring data in larger chunks, and using scatter-gather techniques to optimize transfers over the huge InfiniBand pipe.

All that techno-speak means a couple of really important things. First is raw performance: for applications running on WebLogic, we’re seeing improvements of an order of magnitude or more in both latency and throughput (speed and capacity) for workloads running on Exalogic compared with a comparable typical rack of similar processors, memory, and storage. But not only are the apps faster, they’re more reliable. Exalogic hardware is architected with redundancy everywhere, with no single points of failure. That, coupled with InfiniBand’s guaranteed data delivery and speed, means that any cluster node failure is compensated with extremely fast fail-over for overall high availability.

But the real punch is this: all that speed and reliability aren’t for just specialized, esoteric workloads. Exalogic is designed as a massive consolidation platform, on which you can bring together tens, hundreds, or even thousands of applications from disparate hardware stacks onto a single, fast-to-setup, easy-to-manage platform. And that’s where “Elastic Cloud” comes in—you can use the dynamic cluster sizing capabilities of WebLogic Server and Coherence in conjunction with Enterprise Manager to completely automate application capacity adjustment, forming the basis of a cloud architecture in your own data center.

There’s plenty more to gush about with Exalogic, but that’s a start…

Comments:

Post a Comment:
  • HTML Syntax: NOT allowed
About

This blog covers the concepts, architecture, practices, technologies, and products that provide foundational infrastructure for the cloud.

Search

Archives
« March 2015
SunMonTueWedThuFriSat
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
    
       
Today