SMB Storage Replication: Acceleration VS More Bandwidth

With the cost of secondary storage and bandwidth plummeting, does it really make sense to invest in expensive acceleration solutions to service your storage replication needs? Faced with tight capital budgets, small organizations in particular should consider a "backward thinking" approach. Today, you can get 5TB of Amazon S3 storage for $150/month, and for another $100, you can get a nice fat business class internet pipe to ship your storage there. Compare that annual operational cost with the capital cost of 5TB of on-site storage and WAN acceleration, and you have a tough cost vs. benefit decision to make. 

Much to the dismay of every CFO, IT execs tend to be innovators and forward thinkers. It's not that CFO's despise innovation, they just despise what it costs them. When it comes to large scale WAN Acceleration, it's often going to cost you a lot. On an extremely large scale, WAN acceleration can often pay you back relatively quickly. However, when cash is an issue, and when capital budgets are tight, IT Managers might fare better thinking backward than forward.

It used to be that only large organizations would build DR sites or do storage replication for backup and/or redundancy. Today, even small organizations are demanding some form of highly available storage redundancy. Unfortunately, most SMBs lack the infrastructure and budget needed to do it the way their big business peers do.

Typically, most medium to large sized shops multiplex their storage replication traffic with other critical network services over one large pipe. In this scenario, acceleration can be a huge benefit (the forward thinking solution), but it's not the only approach, and acceleration is certainly not the cheapest solution.

If you're an SMB, think backward. For example, ask yourself if queuing is an option. Depending on the SAN you've invested in, you can often bandwidth throttle storage replication traffic at the SAN level. If you can do deduplication at the same time, even better. If your SAN can't bandwidth throttle the replication traffic, do it at the router level between SAN A and SAN B. If you're replicating into cloud storage, then implement a bandwidth quota on your firewall or internet router. Assuming you have some spare bandwidth to play with, you'd be surprised at how effective this approach can be.If bandwidth over the existing link is too saturated, then you need to make that tough decision regarding whether or not you should try to squeak out more out of the bandwidth you have via acceleration vs. adding more bandwidth. If you step back and really think about your transport needs, you may determine that you don't necessarily need an OC3 from a Tier-1 provider at both ends of the link.

Realistically, you can get tremendous value and lots of bandwidth from a business class cable modem at both ends of the link. By building a VPN and routing your replication traffic over the cheaper bandwidth, you're not only taking the burden off your primary link, but you're now able to replicate storage traffic faster and in real time if need be.

With top tier solutions easily running into the five figure range, it makes good sense to avoid falling in love with WAN acceleration at first sight, especially when it comes to off-site storage acceleration. If bandwidth between site A and B are prohibitively expensive, or you need lighting fast application acceleration at the same time, then all bets are off and you should shop for the right appliance. However, for simple storage replication alone, why not consider other options?