Storage on tap - Cover Story - Network Magazine India

Tackling issues like interoperability and scalability is a routine thing for CIOs. But what really matters today is the availability of information. by Prashant L Rao and Abhinav Singh.

One would think that cost or the shortage of qualified storage administrators would keep CIOs awake at night. Interestingly, it's neither of these factors that keeps CIOs on their toes. It's interoperability that Indian CIOs cite as the biggest problem they face when it comes to deploying storage in their organizations. Next comes scalability. Here there are tools and technologies to tackle this issue head on. Which is why you find storage deployments of dozens of terabytes scaling upwards. Storage infrastructure outsourcing has begun but most companies (particularly in BFSI) are prudent about putting their crown jewels, the core data of their systems, onto a third-party storage boxes. Lastly, while Indian organizations don't have any regulations as such that compel them to act in a particular way, when it comes to archival and storage management, guidelines and ties with US companies are prodding many companies into archiving their historical data in a planned manner.

Manoj Chugh, President - India & SAARC, EMC Corporation says, "Ten years ago companies invested in computerizing different locations—manufacturing in semi-rural areas and marketing offices in cities. Some manufacturing companies centralized their ERP systems, others didn't. The net result of all this is silos of information. There are multiple computing platforms—SIAS, RISC and mainframes—each with its own storage. Huge investments have been made into internal and external storage—NAS and SAN. Today billing information is required for CRM, data warehousing and business intelligence. CIOs need to consolidate and provide access to build competitive advantage for their organizations. It is no longer about land, factories or manpower. Today it is the ability of an organization to harness information that counts. Storage is extremely important as information lives in storage."

The Indian storage market was worth $100 million in revenue terms, for the first three quarters of 2003. About 3,000 TB was sold during this period. 70 percent of this figure consisted of external storage with the remainder being internal storage. IDC believes that the Indian storage market will grow at a CAGR of 65 percent up to 2007. Telecom, BFSI, manufacturing, and BPO are expected to be the biggest spenders on storage this year as are SMEs.

IT Managers usually cite cost, interoperability, and scalability as the biggest challenges when it comes to managing storage. Let's take a look at how they cope.

Interestingly cost is not rated highly as a parameter by most CIOs. Most of the CIOs we talked to rated cost as the least important parameter when it comes to storage infrastructure and the management challenges associated with it.

In stark contrast to the global shortage of qualified storage administrators, Indian CIOs unanimously concur that there is no dearth of HR talent when it comes to storage administration.

Choosing the right tools and technology for storage management is a challenge for IT heads. They have to pick tools that provide an effective ROI.

This is perhaps the most crucial aspect of storage management. "CIOs can change servers but it is difficult to migrate and share information from one platform to another. If you have information in a Xeon server or mainframe, how do you move the information and share it and ensure integrity? Interoperability is extremely important," says Chugh.

"Interoperability has the maximum impact in storage management and data replication," concurs Sandeep Dutta, Director - Strategic Partnerships & Marketing, Network Appliance India Pvt. Ltd.

Virtualization makes it easy to shift resources, particularly when a company already has a SAN in place. IBM's TotalStorage SAN Volume Controller (SVC) gives a virtualized view of storage to all servers. "It can be a Linux box or even a blade on the SAN switch," says Shailesh Agarwal, Country Manager-Storage, IBM India. IBM also has Tivoli Storage Resource Manager, a solution that gives an administrator a view of storage utilization as a resource in the company.

Cisco believes that fabric virtualization is the answer. Here the storage host doesn't have to worry about where the data resides. It has teamed up with Veritas and IBM and it has implemented IBM's SVC software as a virtualization blade for its Director switches.

Idris T. Vasi, Director, Optical & Storage Networking, Cisco Systems, Asia-Pacific says, "Large enterprises with storage boxes from multiple vendors that do not talk to each other end up with only 50 percent utilization as the boxes can't share their capacity. Fabric virtualization gives an organization a single pool of storage."

Benny Chan, Business Development Manager - Storage Networking, Cisco Systems Asia Pacific explains, "Say you have an EMC and a IBM box. You just define a LUN (logical unit number) in both the EMC and the IBM arrays. Fabric Virtualization shows them as one VLUN (Virtual LUN) to the host server."

"Business Continuity is not a technology issue. It's about an organization's requirement for a particular continuity level. A stock exchange has a different requirement than a finance company or a manufacturer. Some companies have a higher dependence on their data than others. Payroll is not critical (throughout the month). So it doesn't make sense to put it along with mission critical applications on a storage array with Fiber Channel disks. But if you add cheaper boxes the setup becomes more complex to manage. If the payroll application has to run for two days a month it can temporarily be shifted to the high performance box and shifted back to a low-cost storage box after that period," says Agarwal.

For IDBI Bank, selecting different solutions was a challenge while setting up a DR site. Defining data retrieval considerations depending on criticality is another issue.

Additional business applications such as warehouse management, SCM, workflow data pushed Hero Honda to look into networked storage (SAN) for business continuity and disaster recovery. The company has a SAN architecture and uses IBM ESS SAN boxes, Fiber Channel Brocade switches. IBM's Tivoli network management software is used for automating tape backups eliminating human intervention for managing backups and shrinking the backup window.

"The rapid growth of data is a challenge as we are expanding rapidly. For instance, we had 30 branches in the beginning of 2004. This has increased to 40 branches in just three months. We are expecting that our stock points will increase from 40 to 65 by end 2004 and area offices are expected to increase from 64 today to about 100 by the end of this year generating huge amounts of data. Managing and securing it will be a big challenge for us," says Daya Prakash, project manager (LG Electronics India - ERP & eBiz) at LG Soft India Pvt Ltd.

Enormous increases in the amounts of data are the norm rather than the exception. For instance, last year Wipro Technologies got 13 lakh e-mail messages every week. This year that's shot up to 23 lakh messages per week.

After going live with its mySAP ERP system in 2001, Hero Honda saw a rapid growth in its data storage requirements. S.R. Balasubramanian, vice president-IS, Hero Honda Motors Ltd says, "Earlier we were using disk-based storage that resulted in poor performance and under-utilization of disks. Additionally, in the event of an ERP crash, both the ERP system and the data needed to be recovered and restored. This meant that no data is available for the next 24 to 30 hours for both our suppliers and for us."

As IDBI Bank uses OLTP software, it uses IO-SAN to get good throughput on its storage network. The bank has consolidated all its business critical applications and its storage boxes are all fiber-based.

"We never outsource our storage requirement needs to any third-party storage service vendor due to security concerns," says Sharma of IDBI Bank.

Prakash of LG states that LG hasn't outsourced its storage needs to a third-party as its storage requirements are growing at a fast pace and it is still not convinced as to whether there is a storage service provider that can keep pace. "We have 40 branches, 70 area offices and 40 stock points, which generate huge chunk of data. We are also apprehensiv e about outsourcing our storage requirements due to security implications," says Prakash. LGEIL has strict corporate guidelines and it is moving towards BS-7799 standardization. After achieving that it may explore outsourcing storage to a third-party.

Wipro Technologies outsources its storage backup to storage service providers (SSPs) in the United States and Europe, as it is cost-effective when compared to doing it in-house. The company has users located across the US and Europe and it isn't practical to set up infrastructure at all these locations. Management of storage backup is best met by outsourcing to a third-party storage service provider. Wipro Technologies may consider outsourcing its Indian storage needs to a third-party provider, but to date it hasn't come across any SSP that can provide the kind of services it wants. Wipro Technologies complies with security standards such as BS-7799 and all security checkpoints are adhered to before storage needs are outsourced.

In the case of HDFC Bank, storage is owned by the bank but it does outsource SAN lifecycle management. "However we have a team within the Bank that understands the SAN and oversees solutions offered by vendors," says S.R. Balasubramanian, vice-president - IT of HDFC Bank.

Hero Honda does not believe in outsourcing its storage infrastructure to third-party service providers. It has built up an IT team and acquired domain expertise in deploying and managing IT systems. "We have a dedicated six member IT team that takes care of managing servers and networks," says Balasubramanian, Hero Honda. The company does not have any plans to outsource its storage infrastructure in the future.

Sharad Sanghi, Managing Director and Chief Executive Officer, NetMagic says, "Though most BFSI companies have concerns about outsourcing data, they do outsource messaging and application hosting." The BFSI sector tends to manage critical data in-house.

"Several banks are our customers for services such as remote infrastructure management, messaging, basic customer facing applications, mailers and statements. We also do strategic consulting, we've done some for ICICI Bank. 90 percent of BFSI companies are more comfortable doing it in-house. We host Morgan Stanley but their core data is not with us," says Sanghi.

NetMagic hosts BPO, logistics, manufacturing and e-commerce companies (Baazee.com is a customer).

Avinash J., President, Hosting Services, Satyam Infoway Limited (Sify) says, "Today most DCs (data centers) make money out of managed services." Sify's managed services customers span the gamut of the major verticals. The list includes the State Bank of India, ING Vysyalife, HDFC Life Insurance, Skoda Auto India Ltd, Indian Oil Corporation Ltd and New Delhi Television.

Sify's main thrust in DR services is in the area of tape backup. Most of its customers want their data to be restored within four hours for which tape backup works well enough. Customers who want faster restoration are offered replication on a storage device (dedicated or shared).

Although Sify's banking customers continue to host their core banking solution in-house, some of them run a backup to their core banking system at Sify's data centers. Many banks host their Internet banking application that links to the core banking piece with Sify.

Most outsourcing contracts are services dedicated to customers. Sify offers hot standbys in alternative locations. Some companies want the locations to be in two different seismic zones, others want them in two different countries. US companies that have applications developed by an Indian partner go for this kind of solution. Applications continue to be hosted out of the US in such cases due to the time delay constraint.

While it is technically possible to use a load balancing solution and do replication online, few organizations go for this as it's expensive. Some large enterprises and gaming companies do go in for real-time replication. Transaction processing, core banking and similar applications may need real-time replication.

Structured Vs unstructured storage

While there are elegant ways of managing structured data including HSM and automated retrieval, unstructured data is a problem. E-mail is a classic example of unstructured data. MPEG and MP3 files are other examples of non-structured data. "Solutions, especially in e-mail archiving, help you archive and manage e-mail to increase productivity. Most e-mail implementations try to restrict usage to x MB per user and ban attachments. Backup and recovery is undertaken through PST files. This is cumbersome and it inhibits user productivity and increasing management costs," says V. Vivekanand, sales director-India, Hitachi Data Systems.

There are various solutions for managing e-mail including Unlimited Mailbox from Hitachi that hashes every single mail based on user, sender, attachments and content before archiving it. The system administrator can retrieve individual files based on parameters or any hash functions. With e-mail becoming a statutory requirement under storage compliance most storage vendors have a solution for this problem. Sun has its Infinite mailbox that helps automatically archive e-mail.