The  move toward widespread virtualization is usually accompanied by new systems architecture. This would address load balancing, failover and high availability, and also support WAN-based access for remote offices and workers (e.g., sales people who work from home).   Because it's designed for efficiently replicating data from one site to another, this architecture streamlines disaster recovery.

Why is this? The data in a virtualized environment usually encompasses virtual machines (VMs), a configuration database and user data. The size of this data depends on your environment – whether it’s desktop or server virtualization. Beyond that, other components of the virtualized environment, such as hypervisors and connection brokers can be easily put in place to accommodate the replicated data from the main site to a DR site. 

Other issues you need to consider when replicating data to your DR site include: network bandwidth, size of your data files and deciding if your DR environment will be a hot or cold site.   Your choice of operating systems makes little difference in terms of DR, because the operating system is part of the overall backed-up data that is sent to a DR site.

In a “traditional” client-server environment without virtualization, backups are usually only done for the actual user data (especially if the data is on a network share) and not for operating systems or applications.  And if your end user devices are PCs, backing up individual PC operating system environments, including individual user customization can be daunting if not impossible.  

Even if your IT department keeps a golden image of the applications loaded on end-user PCs, it typically doesn’t include  customizations.  This is where virtualization can definitely help; each user’s environment is a virtual machine which contains all applications and any customizations (assuming the user is using a persistent desktop). 

From the standpoint of server virtualization, a complete back-up of your server VMs should be easier to execute because the VMs will likely reside on some type of centralize storage.  Most storage systems support storage replication which makes this really easy to accomplish for back up proposes.   The replicated data can be easily sent off site via high speed network to a DR site. 

One of the problems I’ve  seen with sending large amounts of data across the network is the time it takes – some sites don’t have the network bandwidth to handle the data.  This is where WAN optimization becomes your friend! WAN  optimization will let you send large amounts of data across potentially long distances by applying compression, de-duplication and data packet optimization features regardless of the transport protocol or latency sensitivity. This boosts WAN performance and reduces costs by optimizing existing bandwidth. Keep in mind different WAN optimization solutions work better with certain types of data, it's important to do your homework – and perhaps a pilot – before you invest in this technology.

Finally, you need to think about how you configure end user systems to access your DR site if your main site goes down or becomes inaccessible to your staff. 

Will You Be Deploying the Longines Symphonette or the RCA Console with Hi-Fi Stereo in Your Data Center This Year?

Much has been made recently of the ongoing battle between Cisco and HP as they stake out a presence in the emerging market for consolidated infrastructure components. Cisco is touting its Unified Computing platform as a “virtual data center” and “ladder to the cloud,” while HP has introduced the Blade System Matrix that provides a “cloud in a box.”

Buzz-worthy indeed, yet the way forward for both Cisco and HP is a bit confusing, cloudy perhaps.

I have tremendous respect for John Chambers and his ability to continuously grow Cisco's business. Cisco, like HP, is now increasing its focus on products and services for markets beyond the enterprise. A growing portion of Cisco's revenue is coming from video conferencing and consumer products like Linksys routers, set-top boxes and Flip cameras... Yes you too can now experience a “Digital Crib” from Cisco. A broader revenue base is good for Cisco stockholders and maybe for enterprises that have made large commitments to Cisco and need to be confident about their future.

Likewise, HP continues to grow revenue outside of the data center with a heavy focus on consumer PCs, laptops, netbooks and printers, as well as specialized printers for commercial and industrial graphic arts production (see: http://h10088.www1.hp.com/cda/gap/display/main/index.jsp?zn=gap&cp=20000_4041_100 for a look at how graphics created with an HP Scitex printer were used to wrap a train, yes a train, to promote Disney's A Christmas Carol.

Against this backdrop of diversification, Cisco and HP are betting we'll embrace the concept of single, consolidated infrastructure for enterprise-class data centers.

Not so fast…
The Technical Architects at SAI spend a significant amount of time analyzing and debating the emerging technologies and products offered by our alliance partners and other industry leaders. As our team has looked at the “use cases” for consolidated infrastructure, they have raised questions about the implications of deploying components that are “in between” the current model of dedicated components and a future that may be populated with purpose-built IT appliances.

And here's where the Longines Symphonette (LS) comes in. Yes, back in the 1960s and 70s consumers were offered interesting combinations of TVs and stereo components in non-descript faux wooden cabinets – perfect for the easy listening sounds of the LS. These systems initially met with acceptance in the marketplace due to their positioning as a piece of furniture that captured all of your entertainment needs.

This did not last. In short order the quality of home audio components improved, new technologies were introduced (eight tracks and audio cassettes) and musical tastes evolved – how about the first time you heard Led Zeppelin’s Whole Lotta Love. Consoles lost their appeal as purist disdain for moderate quality performance in each area lead to the eventual dis-integration of the components.

During the last two years, we have seen large enterprises make fast and significant progress ( increasing infrastructure utilization rates and decreasing capital costs and operating expenses) with faster and cheaper servers and storage coupled with virtualization tools. Against that backdrop, it's hard to imagine that the current crop of consolidated infrastructure components will gain ground. As these devices evolve to purpose built appliances, the dynamics may change. For now, best of breed individual infrastructure components appear to be a lower cost and higher performance solution.