Infrastructure at your Service

David Hueber

OOW Day 2: Next generation Data Center Management & OEL future

Oracle OpenWorld’s day 2 is now over … with some rain… 😥  We had some interesting sessions such as Fusion Middleware, RMAN or NoSQL.  😀

In this OOW ‘Road Journal’ post, I am going to focus on two of them, introducing a coming Data Center Management solution and giving an outlook on the Oracle Enterprise Linux roadmap.

Trellis: a brand new DCIM platform

Trellis is developed by Emerson Network Power, who based it on a simple observation of the gap between Facilities Management and IT Management. For Emerson, a Data Center has a 3 level structure:

  1. Physical Layer
  2. IT infrastructure
  3. Application layer

In the current Data Center management view, the physical layer is managed by Facilities Management, while the two others are under the responsibility of IT Management. The different tools used by each of them and the complexity increase in the Data Centers ecosystems, especially with Cloud Computing, is bringing this solution to its limits. In addition, the balance between efficiency, availability and capacity is not easy to maintain and leads to increasing energy requirements.

To match these challenges, Emerson works on a new Data Center Infrastructure Management  (DCIM) platform: Trellis. The concept of Trellis is based on the triptych:

  • See
  • Decide
  • Act

The cornerstone of Trellis is on the See stack, with the development of an “Universal Management Gateway”.

umg

This box aims to gather data from heterogeneous “probes”, going from the cooling system to the server usages (CPU, Memory etc.) and running a pre-processing to send only relevant data to the platform. This data is then presented and analyzed thanks to a UI based on Oracle Fusion Middleware, which offers several functionalities like Inventory Management, Virtual Integration Manager, Change Planner or Site Manager.
A cool function I saw there is the possibility to get physical or even electrical diagrams from the whole Data Center and the ability to drill down a to single server / rack.

Trellis will be able for private as well as public cloud deployment. Release 1 is planned to come out on 1 March 2012. More info can be found on www.emersonnetworkpower.com/trellis

Oracle Enterprise Linux, what’s next

As Oracle Enterprise Linux (OEL) is celebrating its fifth anniversary, this solution is really become a reliable platform for Oracle databases implementation. OEL is already used by 8000+ registered customers. I for myself recommend and use it at dbi services. Now with the new Solaris 11 that just came out, it is interesting to have a look on the OEL future and positioning by Oracle.

Oracle Enterprise Linux Strategy

The OEL strategy presentation pointed especially the following out:

  • OEL still is the basic development platform for Oracle products.
    This means that any package for Linux is developed, tested and validated on OEL.
  • The full Red Hat compatibility will remain
  • OEL is supported by dedicated teams (developments, Support, Q/A)
  • Its target is to be the best performing Linux platform.
    Therefore OEL is tested and validated on the Exadata/Exalogic platform, one of the strongest machines on the market.
  • It provides implementated Best Practices. Here a interesting point! Oracle provides 100+ validated/qualified configuration with the corresponding documentation and best practices. Have a look on http://linux.oracle.com/pls/apex/f?p=102:1:3854354817198153

Oracle Enterprise Linux status

The latest released of OEL is the 6.1 with the Unbreakable Enterprise Kernel (UEK) release 1, which is based on Linux Kernel 2.6.32 plus some optimizations and bug fixes. Beside the version 6, the version 5 release 6 is still available.

In the current version, UEK supports:

  • Up to 4096 cores and 2TB Memory (Exa-tech oriented)
  • T10 Data Integrity standard (see www.t10.org), which guarantees integrity/corruption check on write level instead of read on

In addition, an interesting point is that Oracle clusterware and Oracle Linux Management Pack (for Enterprise Manager integration) are free for support customers. DBAs can also find a real interest in the Oracle validated RPM, which offers a huge improvement in Linux server preparation for Oracle database installation.

Finally, OEL has recently introduced Kspice, a “yum-like” tool allowing for online and zero-downtime Kernel upgrade. Kspice runs as a client on each OEL box and connects to Kpice’s repositories to upgrade the OEL’s kernel. This all works without application nor I/O downtime. Furthermore Kspice allows easy upgrade rollback. We could even imagine some kerne temporary upgrades in case of incident or bug investigation, which can be rolled back once done.

Oracle Enterprise Linux roadmap

Here is the announced focus of development for the next releases:

  • Dtrace integration
    Cool tool coming from solaris world featuring server overview but also tracing programs or probing low level information
  • T10 Data Integrity
  • OCFS2
  • Infiniband (especially for Exa technologies)
  • Transcendent Memory
  • FedFS
  • Xen
    The root of Oracle Virtualization solution, Oracle VM
  • Linux containers
    A kind of Solaris Zones principle applied to oracle Enterprise Linux

In addition, the OEL team is targeting a 12 to 18 months refresh period for the Unbreakable Enterprise Kernel, with a release 2 based on Linux kernel 2.6.39.

See you tomorrow for more news on OOW 2011!

6 Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

David Hueber
David Hueber

Chief Executive Officer (CEO), Principal Consultant