by August 4, 2004 0 comments

Storage and disaster recovery held the limelight for the last two years. But this year has been rather slow on the storage hype front, with vendors moving on to fresher pastures. So, perhaps it is a good time to sit back and take stock of where we have reached. The great storage push started with the debate on NAS over DAS (that is Network Attached Storage over Direct Attached Storage for those who have already forgotten), which soon enough became an extended debate of DAS vs NAS vs

While NAS vendors may claim victory for their line of argument because of an increase in numbers for NAS devices, the fact is that, Direct Attached Storage continues to hold a significant percentage of IT hardware budgets. On the SAN front, the big news last year was supposed to be iSCSI. iSCSI was supposed to make SANs cheaper and as a result, speed up adoption. A year down the line, SAN adoption is still limited to the big IT implementers. Why? The answer to this question will yield quite a few insights into which way data storage is headed. The concept of SAN assumes that you either have a humungous amount of data or that you will be able (and willing) to consolidate across the enterprise. And finally, it assumes that you are willing to put down fairly large amounts of money required for an implementation.

Tape in its various forms is still emperor of backups. Rumors of the imminent demise of tape continue as before, as disk-based backup systems start making their mark
Storage Management:
Storage management: Like in other areas of your IT infrastructure, management is the current mantra for storage also. Storage management includes in it, concepts such as ILM and DR
San: iSCSI and IP SAN were supposed to make the SAN commonplace. While prices have come down, that dream is yet to materialize. But the chances that the SAN will be part of your consideration set when planning your next IT budget is fairly high
Appliances: Appliances made a big buzz some time back. While no longer hogging the limelight, appliances are slowly making their way into the storage infrastructure

Users tend to be protective of their data, and as many a CIO has learnt the hard way, it may not be easy to convince them to store it ‘far away’ from them. Storage consolidation also requires fast access and robust storage-management infrastructure and practices to succeed. Fast access means that like with enterprise and voice applications, you need to be able to assure a quality of service level for data access also.

Discussions on storage consolidation led to another industry buzzword last year, ILM (Information Lifecycle Management). Simply put, ILM states that the value associated with information changes with time, and hence, enterprises can plan to move older (and hence less valuable) information to cheaper devices, as they can afford larger access times there. Like with most buzzwords in storage, this one also has had only a limited impact, only in the upper realms of the user market. The reasons again are pretty much similar to what happened with enterprise applications-complexity of implementation, high costs, interoperability issues and finally internal data issues in enterprises together have played spoil sport. As a concept, ILM does make sense, and most enterprises have been practicing some form of the same for ages. Hopefully, the issues can be resolved quickly enough.

With increasing data volumes, the load on backup systems is on the increase. As the quantity of data being committed to backup systems rise, the backup window required also increases. As a first step, most organizations will invest in faster and faster tape technologies. In spite of such investments, the backup window is likely to keep widening relentlessly, leading to the search for alternates to tape.

Hard disks are emerging as a potential competition to tape in this segment. That hard disks could be used for backup is not a new concept. What was holding it back was the much higher cost per GB of disk, when compared to tape. Now, with both disk prices and capacities approaching tape levels, vendors and users are using disk as a backup option. Last year saw a slew of disk-based backup solutions make their appearance in the market. This has been both, as backup appliances from vendors such as Overland and as full-fledged enterprise backup systems from the likes of EMC. Disk-based backup is not a simple disk-to-disk copy. Rather, the disk-based device is made to mimic standard tape libraries. This one assumes is a tribute to the stability and standardization (and hence interoperability) that tape-based backup systems have achieved over the years.

Talking of stability brings us to one of the foremost concerns with backups. Will they really work when we need them? It is well and good to religiously keep taking backups, automating them, sending them offsite and all that. But will they work, when we need them to? With the increasing backup loads this is one crucial element that most of us seem to ignore. Rigorous processes need to be put in place not only for creating backups, but also for testing them. Another tendency is to take backup media for granted. The media has a certain lifecycle. Using them beyond the planned lifespan is as dangerous as not having a backup. Question is, how many of us realize this? And more importantly, how many of us convert the realization into actionable plans?

Another concept that briefly made its pitch was storage outsourcing. This involves outsourcing the entire storage infrastructure and its management to a third party datacenter. Only a few have tried this approach, and as yet, it is too early to pass a verdict on the efficiencies of the practice.

In short, the perspective on storage and backup technologies continue to be mixed with no single solution emerging as a clearly preferred choice, be it technology or be it product type, compared to others. Strangely enough, this was not the scene a couple of years back, when things were more settled and when the hottest debate was about making a choice between LTO and AIT. In the short run expect a churn of technologies and products, even as data storage and backup requirements keep growing by the day. 

By Krishna Kumar


No, it is not about appointing maids to take care of your storage and backup systems. MAID stands for Massive Array of Inactive Disks, a new attempt to using inexpensive hard disks in the place of tape for backups. MAID systems use hard disks that are powered on only when required. The disks are specially designed to be able to handle the repeated power on-off cycles. The concept is being spearheaded by Copan Systems, who have announced the Revolution 200 T based on MAID technology in April this year. 

The Revolution 200 T scales up from 56 TB all the way to 224 TB, with up to 896, 2 GB SATA disks. Physically, the device is a single cabinet of up to eight shelves, with 112 drives per shelf, loaded in eight canisters per shelf, with 14 drives loaded per canister. It can provide a data throughput of 2.4 TB/hour, using multiple 2 GB fibre channel interfaces. It emulates tape drives and libraries from multiple vendors. The system is priced at $3.5 per GB

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.