SAN based Remote Replication – Changing the Paradigm

Not all SANs are created equal when comparing remote replication capability. It’s amazing to see all the storage vendors marketing their capabilities as if they were technology leaders, when in fact most of their offerings are feature deficient or very expensive to deploy. Let’s set the record straight by taking a look at customer requirements.

I’ve talked with many customers that require offsite replication for DR/business continuity. Here is their consolidated “critical requirements” wish list.

Asynchronous Local and Remote Replication:

1. Scheduled SAN based asynchronous replication over any distance using standard IP networking gear, meaning no Fibre channel to IP bridge equipment.
2. Must work over almost any link bandwidth (T1 being the most common).
3. Technology should minimize the replicated data rate by sending only block level changes.
4. Must have some bandwidth management tools to slice up link bandwidth and share it with other applications that use the remote link.
5. Must provide the capability of failing over servers to the remote site, failing back and incrementally resynchronizing the primary site with the changes.
6. Should provide the tools necessary to replicate application-consistent data sets, making application recovery a simple matter of mounting a set of consistent volumes.

Synchronous Local and Remote Replication:

1. SAN based synchronous replication using standard IP networking gear, meaning no Fibre Channel to IP bridge equipment.
2. Should provide seamless application server failover and failback to the remote site and back with no manual intervention and with incremental data re-synchronization.
3. If a site fails, volumes must automatically remain online at the remaining site.
4. Must work seamlessly with server and application failover software.
5. Must work seamlessly with server virtualization HA products.
6. Capable of utilizing the bandwidth to of fat pipes (parallelism).
7. Must protect against “split brains” (insures data integrity at both sites when the link between sites is down and transactions are still being processed).
8. Capability to bring a site down for maintenance without downtime.
9. Must support applications running on more than one site and replicating to each other.
10. Geographic/site awareness with I/O preferencing based on site/subnet.

When you start going down the list of SAN storage vendors, checking off those that do not meet these basic critical requirements, your options start dropping off rapidly. When you add a requirement to keep costs down, the list thins out even more. Most of the mid-range vendors only support SAN based replication using Fibre Channel, which means expensive Fibre Channel to IP bridge equipment is required on both ends of the pipe along with training and professional services to install, not to mention the outrageous software licensing costs.

There is only one iSCSI vendor out there that offers a fully featured SAN based asynchronous and synchronous replication solution over IP, and that’s LeftHand Networks. Not only does LeftHand steal the show in this regard, but most of the features come standard in the base software offering, which is truly an anomaly, and I challenge any SAN vendor out there to prove me wrong.

If you are looking for a low cost, SAN-based remote replication solution of just about any flavor, over just about any pipe, that leverages your existing Ethernet and IP networks, and works seamlessly with your applications and VMware’s HA capabilities, give LeftHand a call. We’ve got the BC/DR solution for you.