| By Dmitry Sotnikov |
Article Rating: |
|
| December 1, 2008 07:35 AM EST |
Reads: |
7,988 |
DAY ONE
This week I am at the moderately sunny and warm (can’t believe I was driving through a snow storm in Detroit just 3 days ago) San Jose at SYS-CON’s Cloud Computing Expo.
I intent to blog from here all three days - so if you could not make it to the event you can at least get the gist of it from cloudenterprise.info.
The history behind the event is pretty straight-forward. SYS-CON has been holding their International SOA World Conference for ages. As the hype behind SOA started to fade, they did what any company would do: try to diversify. Which led to this 14th SOA World becoming also the 4th International Virtualization Conference, and the 1st Cloud Computing Expo.
There’s still a lot of SOA content - especially in the general sessions - but, hey, we all know that SOA and Cloud Computing are not that far off each other. The difference is that SOA is pushed by consultants who are scaring everyone with talks about how everything needs to be re-architected, and Cloud guys are vendors like Amazon with solutions they want to be relatively easy and ready-to-use right away. And this seems to be making a lot of difference!
Anyways, this conference turned out to be a fairly big event with 3 days packed with more than 90 sessions in 7 tracks.
The conference kicked off with an SOA keynote. David Linthicum (SOA consultant and Infoworld SOA podcast host) tried to downplay the Gartner’s report on SOA disillusionment and link SOA to cloud computing (if you have SOA taking pieces of your application to the cloud is going to be easier), and then went on to some scary architecture diagrams.
After another general session on SOA we could finally get to the cloud tracks. Here are my notes from the sessions I attended.
Stuart Charlton from Elastra suggested his classification of cloud platform architectures and announced their upcoming (later this year) public release of markup languages to describe models and policies for software systems deployed into the cloud. Their goal is to bridge the gap between applications and infrastructure by providing what Stuart called Architecture-Aware Clouds. He also mentioned the upcoming support of Amazon and VMware, as well as their efforts to solve the licensing issues for the platforms used in their systems.
[Download Stuart's slides]
Reuven Cohen - Enomaly - turned his session into an open discussion on the cloud and the hurdles in cloud computing adoption. The issues mentioned included:
- Regulations not allowing data to be hosted outside the country in which a company operates,
- Data and application portability to avoid vendor lock-in,
- Guaranteed data destruction when no longer needed by the client,
- Failovers between datacenters: external belonging to different vendors and internal on client’s premise,
- Existing enterprise apps and how they would need to be re-architected for cloud scalability,
- Certificate not revocation issues (e.g. am employee leaves the company and you need to revoke certificates - with Amazon that would kill your state, Nirvanix seems to allow you to handle that gracefully),
- Amazon entering the content-delivery market (providing local replicas of your media around the globe for low latency access) with very competitive pricing, and Rackspace is partnering with LimeLight networks to also have a solution in the area.
Kevin Haar - Appistry - was talking about how applications are now “all that matters”, and how proper architecture gives you access to any type of cloud: public, virtual private, or private.
Patrick Harr from Nirvanix talked about cloud storage and how it should become more cross-platform/cross-vendor, allowing customers to move data, form cross-vendor failover clusters and so on.
[Download Patrick's slides]
Finally Billy Marshall from rPath - pitched the advantages of using their company’s virtual appliances in cloud computing. Basically, if you create your application machines as rPath appliances you can then save them to the machine format you need: VMware, or Hyper-V, or XEN, or KVM, or Amazon - and deploy anywhere you want. Plus, you get their versioning technology and incredibly small machine size. The biggest drawback is that they don’t provide anything beyond individual machines so if you need failover, load-balancing, queuing service, databases and so on - you are on your own implementing that as individual appliances. It was interesting to see them already having Amazon integration right in their online appliance management UI - with more platforms to be integrated in there next year.
[Download Billy's slides]
Subscribe to JAVA Developer's Journal Email News Alerts
Subscribe via RSS
