More Content - Including Podcasts

Thursday, June 9, 2011

HP Discover and The impact of cloud services on disaster recovery

Another potentially interesting session with the second least desirable timeslot. Least would would be tomorrow at 08:00.

Julien Furioli from PWC presented, he is the architecture and DR practice lead. He assured us that the content here is from the field, and not based on studies or external research groups.

The focus is on financial and retail industries, the challenges and opportunities for cloud and DR and BCP.

Cloud based opportunities are not measured solely on financial benefits, but the high availability and flexibility offered. Challenges in the adoption of cloud computing are security not being considered mature from the client perspective, potential hidden costs in maintaining similar configuration between prod and DR sites, including licensing costs, and determining the correct model for your organization, public, private, or hybrid.

Comparing Financial services to Retail industries as a whole, the DR maturity is much higher in FS. Most retailers aren't consider low RTO and RPOs such as FS organizations who seek specific metrics and measure and test frequently. A large reason is the regulation on the FS industries which is not as much an impact on the FS industry, where they have a higher tolerance for outages and disruptions. DR awareness overall is low in Retail.

Comparatively, FS industry is mature in the the virtualization field, at 30-50% adoption, where Retail varies much more greatly. Neither industry is moving strongly to the cloud services, with both experimenting lightly with public and private cloud solutions for non main line of business services.

The cloud based DR solution being experimented with by Retailers, is hosting a stateless browsing only replicant of their online presence in a cloud architecture. This won't enable commerce in a DR situation, but at least keeps the face of the organization available.

Some global retailers are planning a move of their complete offerings, by making cloud computing the standard for their compute architecture. They will use two competing cloud offerings in a private cloud configuration and offering failover between them.

The main advantage for them is the seasonal bursts of activity these groups see, and the ability to work with these, as well as reprovisioning cloud resources from Dev/QA to production in advance of known peaks since code freezes go into effect prior to peak seasons.

FS is focused more on using private cloud for VDI and enhancing DR. This allows complete standardization of the desktop globally and a high availability virtual desktop.

Weakness here would still be the network.

Illustrated a diagram noting tier 1 - 3 of DR for HA designing using Cloud computing. Second and third tier architectures are predicted to be most readily adopted.

The message that came next summarized an interesting point, which is that DR people make better use of their time up front architecting a solution rather than rolling up their sleeves and orchestrating details without a failover and recovery infrastructure that aligns with the business needs.

At 80% complete on the design, go to the lines of business leaders, to objectively discuss and understand who's business processes are priorities, to ensure that they will be able to get the RTOs and RPOs they truly need.
I would assume he has assumed we'd do some requirements gathering up front and this is intended as a design gateway checkpoint.

Cloud technologies can provided enhanced HA for lower DR tiered services, and flexibility for peak periods.

Public cloud is liken to an engineering offering, more so than a proxy rises business offering. Clients cannot repackage solutions easily.

Focus is expected over time to shift from managing Dr plans to supporting architecture and engineering. Testing should become more frequent and business centric. Automation needs to replace manual processes to leverage the advantages of cloud services. RTOs should trend toward zero, and RPOs will ultimately become negligible. DR manual processes can be expected to manage the RPO gap and catch up on the data.

Within 5-10 years, expect to see DR teams shift to architecture, QA, and engineering resources.

Another promise of this shift is more emphasis on testing because resources are freed up to do it, and the importance of proper testing grows with increased automation.

Current complexity of the DR orchestration model is based on application and infrastructure complexities inherent in the existing architectures. We should get away from rebuilding applications being considered service resumption. Cloud based solutions bring HA benefits to lower tier virtualized shops.


- Posted using BlogPress from my iPad
Location:Las Vegas, NV

No comments: