Purpose-Built Remote Office Converged Infrastructure
Remote office converged infrastructure (CI) is a relatively new term. It is the idea of using a single or pair of devices to simplify running multiple infrastructure functions at a remote site.
It is different from the ‘traditional’ CI technologies in that those are designed for data center scale-out architectures—most start at 4 nodes and need at least 3 to maintain high availability (HA).
This is overkill for most branches because they rarely run more than a handful of virtual machines (VMs). And none of the current vendors have native network optimization, which is critical for transmitting data on low-bandwidth, high-latency networks.
Many customers have deployed clustered ESXi servers sharing dual controller storage arrays to remote offices. This ‘legacy’ solution has been the best option for the better part of ten years.
This works, but is complex to implement and still needs additional solutions for backup and recovery. If WAN optimization is needed, that’s another appliance to add to the limited space generally available at these sites.
Riverbed SteelFusion directly addresses these challenges and it does it with elegant efficiency. Not only is it possible to utilize a single appliance to provide the entire infrastructure for a remote office—SteelFusion can provide high availability (HA) using only two appliances (Edges). These are mirrored and purpose-built to run ALL of these functions at the branch.
We achieve HA by replicating our BlockStore (more on the BlockStore in a minute) to the 2nd Edge. In the event one Edge fails, the VMs running on the other Edge are unaffected. The VMs on the failed edge are re-booted in minutes (via vSphere HA). And if your application is clustered, there is no loss of availability to the users regardless of which Edge failed.
An Edge failure is rare because all components in each Edge are redundant. Another significant benefit of HA is the ability to vMotion VMs when you need to do maintenance- with no downtime.
Up to this point, I’ve focused mostly on hardware. SteelFusion is much more than this.
Consider the challenges of maintaining storage capacity and consistent backups at even a small number of remote sites. I’ve yet to have an IT leader tell me they are confident that backups are successful every night at every location or that they are content with the 24 hour Recovery Point (RPO) that provides for.
Again, SteelFusion was purpose-built to address these areas. Taking them one at a time
1. Instead of storing 100% of the data of these remote sites locally at the branch- SteelFusion caches the most recently used data in our BlockStore. This means that you no longer have to design for 5 years of growth when deploying a site. The challenge of adding storage capacity at hard-to-reach-sites is bad enough, much less the hardware expense.
2. Consider that the majority of the data at any given site resides on file servers where the data has been accumulating for 15 years and rarely accessed- and you’d be understandably frustrated at this expense.
3. Any consideration of telling users to clean up their data is instantly killed the moment IT is blamed for removing files that are still needed. That’s assuming the person who created that file still works there and remembers it. Any file archiving software comes with its own complexity and I have some horror stories they introduce to file server management.
4. Because SteelFusion is built on the industry leading WAN Optimization platform, files that aren’t in cache are fetched as quickly as possible when requested.
5. Know that SteelFusion does not require high performing storage in the data center. We can use ‘cheap and deep’ storage. Customers either get some extra life out of the slow array nobody wants to use, or just add a shelf of 4TB RAID6 drives. Managing a single storage pool, in the data center, is infinitely easier than doing it at every branch. See the map below for a visual representation of this.
1. SteelFusion was designed from the ground up to keep ‘data in the data center and performance at the branch.’ All changes at the branch are first written to local cache and then WAN optimized back to the data center. It is common for us to take 80% of the data off of the limited WAN pipes with SteelHead technology.
2. Consolidated reporting gives you very detailed information on data at all the remote sites that has yet to make it back to the data center. There’s even an ETA report so you always know what is at risk and how long it will take to get it to the data center.
3. SteelFusion has the ability to use an encrypted Internet tunnel for sending data to the DC, eliminating this data from the expensive MPLS network.
1. Imagine having the ability to recover the most recently used data and most important applications first with 80% deduplication over the network. SteelFusion continuously monitors which files are being accessed and restores these first. There are many layers of granularity for manual recovery as well. All for something that’s only needed when there is a true disaster.
2. Alternatively, for disasters where there is no branch left to recover to- because all of the data is in the data center, simply mount the storage to the production virtualization cluster and boot the VMs. Yes, you’ll likely have to update IP addresses because of the different subnet, but that’s a simple process. When the remote site is restored, you simply map the storage back to the remote office and let SteelFusion do its magic to get the VMs up during a planned window.
In summary, there are other solutions for infrastructure at remote offices, but none are purpose-built for it. Consider the two maps below and the challenges of managing storage capacity and backups at every site versus doing it in the data center.
SteelFusion offers the simplest to administer, fastest to deploy, and most efficient solution possible.
Learn more about SteelFusion.