We are a 300 strong company providing IT services to the government. Recently, we have been thinking about introducing NAS as a mass storage solution to the ever-growing demand for office automation file and e-mail storage space.
We also have various applications running on Informix (both UNX and NT) and MS SQL. My primary concern is that by definition NAS does not support block-level access that is a requirement for some databases such as Informix. However it seems that some manufacturers are offering appliances capable of providing both file-level and block-level access. One such example is the NS3000 from IBM.
So, I'd like to know,
Q1. Are such appliances in real fact well suited for running databases or should we stick to file sharing only and resort to a SAN for the former?
Q2. Does it make sense to replicate a NAS to another NAS in a disaster prevention and recovery scenario? Would you consider FC or Gbit Ethernet as the interconnection between the two appliances?
Q3. As far as Ethernet traffic is concerned, would multiple NICs (on a NAS) connected to individual ports on a switch help in relieving some of the traffic congestion during peak usage?
Thanking you in advance.
I don't believe IBM has an NS3000 product. They have a NAS 300 and a NAS 300G. The block I/O capability you might be referring to is if SANergy is enabled on the NAS 300G. SANergy will give you the semantics of file access but, using software installed on each accessing server, turn that into block I/O over another path to the storage through the SAN. This is done for the performance needed that can't be obtained through network file access. IBM has sold many copies of SANergy.
In answer to your other questions:
A1: Running certain applications such as databases in some environments require performance that may not be obtained with NAS devices. You need to understand your requirements to determine what fits your needs. In addition, some vendors may not support NAS devices for their products. Be sure and check.
A2: Replicating one NAS to another is fine for quick recovery. It does not replace the need for making copies on tape. The technology to use is a business decision. Distance and data rate (translated to the time to do the copy) cost money and you need to understand the business requirements to make an informed decision.
A3: Multiple NICs probably would not relieve any traffic congestion. The congestion is caused by workload and you either need: less work (isolate traffic into different nets) or bigger pipes. If your workload is going to increase (most does), you need to look at your infrastructure and the solution you are choosing. Will it continue to be effective given future requirements? One of the major issues with putting storage on a network is who controls it, a storage administrator or a network administrator? There are sociological differences here pertaining to security, management and control, etc. This can be an intractable problem. Be sure to cover your security issues.
Evaluator Group, Inc.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum.
This was first published in February 2002