Many customers today expect ‘WAN acceleration’ technology (sometimes referred to as WAN Optimization) as part of their MFT vendor’s solution offering. In general this is a useful addition to the MFT feature set, and can certainly reduce file transfer times in a wide variety of scenarios. However, customers should have realistic expectations of what these acceleration technologies can offer, and be cognizant of the limitations and constraints imposed by the carrier network itself.
Customers should question any absolute, unequivocal claims an MFT vendor makes regarding performance improvements achieved using their particular approach. A claim of “7x” or “30x” improvement without any documented caveats is simply not credible. The key point is that observed performance enhancements in the WAN are probabilistic, not deterministic. A file transfer occurring multiple times between the same endpoints will in all likelihood produce different latency measurements depending on a large number of factors:
- Time of day
- Day of week
- Physical media traversed
- Design of intervening switch fabrics and router queues
- SLA agreements with the carrier
- End-to-end QoS provisioning (if any)
- Burstiness (jitter) of co-mingled traffic, etc.
Techniques for improving WAN performance vary by vendor: data caching, compression, truncation, protocol optimization (usually proprietary, as an enhancement to TCP at the transport layer), traffic shaping, and de-duplication, just to name a few. Customers should ask many questions and perform their own “real world” tests to ensure they are in fact receiving the transfer performance improvements they expect, under conditions that are common to their WAN environment.