Answers to Issues in Managing Storage

All Storage is Not Equal



This is essential to decide on your overall storage strategy. To begin with,

all storage is not equal. Look at how information is accessed: the frequency of

access is greater when the data is new and it reduces with the age of the data.

There is cumulative  effect of collecting data. As a result, at any point of time

we have more old data than new data.

What does this imply? First, new or recent data is the one that we use to run

the business and generate revenues. Old data intrinsically has lesser revenue

generation potential. But old or new, data has to be preserved.

Second, the way data is used is dependent on what applications we propose to

use–building a data warehouse to track customer behavior, business

intelligence, and the like. The latency of information or ‘the time we are

willing to wait’ for the data to be available is dependent on the kind of

application.

Bottomline: The frequency of access of data and the latencies required

by the various applications determine how data has to be stored. This is called

‘information lifecycle management’ (ILM). ILM is about understanding the

lifecycle of data and deciding on storage options best suited so that the

overall storage costs are lowered.

CIOs’ Recommendation: Build a storage hierarchy based on the age of

information and latencies required. Storage technologies are available at

various price points; call it a cost hierarchy. Marry the cost hierarchy with

the storage hierarchy to determine the best options.

Optimizing Bandwidth with Efficient Storage



 The point is plain and simple. If you have 20 TB of data and 2 Mbps of

bandwidth, it will take ages to get your back-up or storage done, at least on a

daily basis. But there are ways to handle this. First, try incremental back-ups.

Alok Kumar, director, IT, Tata Teleservices (M) advises this method because it

takes just an additional five or six minutes to back-up the incremental data.

Various packages and tools like Oracle and Veritas allow you to manage

incremental back-ups.

Another option would be to look at a thin-client (like Citrix) kind of an

environment. The same 2 Mbps bandwidth can serve double or triple the number of

users you currently have. Most packages now are Web-based rather than being

client-server based, therefore bandwidth requirements are bound to fall.

Bottomline: Storage consumes bandwidth but one can avoid bandwidth

becoming a bottleneck. This means, bandwidth per se is a non-issue when it comes

to storage. There are practical ways of getting around the issue.

Best Practices

Plan storage as a separate entity. Have a storage policy in place with redundancies built in at various levels. Standardize everything end-to-end. For example, standardize the tape libraries- nobody can buy or use any other than a particular tape library so that inter-portability is very easy. Backup requires moving very large amounts of data from one system to the other. Deploy the best computing resources including a high-end server to do the back-up. Otherwise, you will end up delaying the backup process.

CIOs’ Recommendation: Do what the business requirements demand.

Examine variety of options like incremental back-ups aided by software

replication features provided by vendors like Oracle, IBM, Sybase, Veritas and

the like; island the storage network using separate switches; look for Citrix-type

thin-client computing–or innovate using technology or pure common sense to

fulfill the business needs.