The Future Data Centre

With increasing dependance on Web-based apps and extension of the enterprise to suppliers, OEMs, and third party vendors, it has become essential to redo the data centre says Rahul Neel Mani

Information Lifecycle Management requires true virtualisation horizontally across the storage layer and vertically with the server layer,” says Arun Rawtani

Almost all IT savvy enterprises are moving towards a Web-based computing architecture aligning CRM, SCM and ERP in the extended enterprise. With this, the legacy data centre (DC) approach or building blocks have to be reconsidered. The new DC should be capable of blurring the boundaries or isolated islands of computing by using grids, virtualisation, adaptive enterprise, outsourcing etc. and bring business agility to the extended enterprise. This new DC, by all means, should be such that it can create complex business processes and simplify management with the help of policy based tools such as information lifecycle management (ILM), storage consolidation, Web services etc.

Today’s technology architects agree that a new, Web based computing model is taking shape but unfortunately no one knows what it will look like and be called. Many names are making the rounds - On-demand computing, Grid computing, Utility computing and Adaptive Enterprise to name a few. Actually, each vendor wants to patent its own name for this new data centre initiative, which is nothing but a need based source for need based computing. Be it IBM, HP, Oracle, Sun, EMC, SAP, Microsoft or any other vendor, users have to be careful while going in for these lofty plans. They should carefully study the roadmaps of their key technology partners while taking decisions about all the necessary layers of the data centre and the applications residing in them.

What’s new in the Data Centre?

As opposed to the traditional data centre, where the core-computing environment hosted critical business applications the new data centre is distributed in nature. It has a mix of Web-based applications facilitating the extended enterprise encompassing an enterprise’s wide area network. In this situation, applications and computing resources are made available to users beyond the reach of stand-alone devices. What is of foremost importance in this process is the design, security and management skills of the data centre where applications are built with components and services span the extended enterprise. As the next step – storage, computing, and network resources – all become virtualised and can be used and maintained from anywhere. You must be wondering whether there is any rocket science involved here. No! It’s simply a new physical design assigned to the DC with some new policy based solutions such as ILM, storage virtualisation, Web services etc.

Some new buzzwords that litter the new DC landscape are: utility computing, where computing resources are provided and consumed like utility services such as water or electricity based upon your requirements. Next comes autonomic computing, the new name for self-healing systems. Grid computing is a concept where processing power is optimally utilised by tapping unused resources in the enterprise. Virtualisation involves providing storage, computing and network services as pools of resources to be drawn upon as needed. All these things, or a combination thereof, will make your DC efficient and optimal.

The building blocks

IP-based storage networking will simplify management and lower the total cost of ownership, as well as allow the same network administrators to manage both the local area network (LAN) and the SAN, says Shivaji Chatterjee

In a layered approach, the new data centre has many facets including servers, networked storage devices, network infrastructure, security appliances, application and infrastructure software, mobile infrastructure, network management etc.

l Servers – The heart of the data centre: Before zeroing in on server deployments. It is very important to decide which server architectures best support applications such as ERP, SCM and CRM, and emerging Web-based applications? One has to judiciously choose from the available standard Intel servers, proprietary IBM, Sun, SGI or HP servers, or blades. Adoption of 64-bit systems and a grid-enabled server architecture is on the rise. IBM’s pSeries runs AIX (Unix) and Linux, and it supports 15,000+ applications. Sun’s UltraSPARC machines run Solaris (Unix), support 12,000+ 64bit apps and, of late, support chip multi-threading and dual core. HP’s 64-way IPF and PA-RISC are also making the rounds. The next best thing to these server families is to outsource core processing as an on-demand option or to go a step further and embrace server consolidation.

When we talk of servers, we can’t leave aside the disruption caused by blade servers. A blade server chassis houses a dense cluster of modular servers that can be managed and deployed virtually. Blades are a potentially disruptive technology and as server consolidation happens more network services will go onto them. Vendors such as IBM, HP, Sun, Dell, and Cisco are all gearing up for this evolution.

l Storage: You can’t escape it - This market is the most volatile and is moving at a very fast pace towards network based storage and virtualisation. As a CIO, your most important task is to ascertain what role will SAN, NAS and storage over IP play in your company? Important business issues are also reshaping the storage scene. Says Shivaji Chatterjee, senior director Marketing, HECL, “The relevance of storage over IP in the new age data centre lies in the convergence of network and storage architecture. IP-based storage networking will simplify management and lower the total cost of ownership, as well as allow the same network administrators to manage both the local area network (LAN) and the SAN. Although storage over IP is picking up, it is hindered by the fact that TCP/IP protocol processing utilises a significant portion of the host CPU’s resources and that TCP is a slow protocol for storage functions. In short, IP-based storage will require placing the TCP/IP and iSCSI data movement in the hardware.”

Event correlation technologies that are becoming common now would enable the console to make intelligent decisions and take proactive action to enforce security and compliance policies, says Felix Mohan

Arun Rawtani, Country Technology Solutions Manager, EMC India & SAARC says that the best way for an enterprise is to have scalable storage solutions that are interoperable with all kinds of servers and can operate in an heterogeneous environment. Storage virtualisation allows a business to pool storage sourced from different vendors, which helps system or storage administrators to control far more storage more easily and with less margin for error than would otherwise be possible. “Virtualisation is the next frontier of automation. ILM encompasses everything, right from the application down to the storage level. Automated network storage is the first step towards Information Lifecycle Management. ILM requires true virtualisation horizontally—across the storage layer—and vertically with the server layer,” says Rawtani. If today’s enterprises are really keen on using storage properly and effectively, ILM is the key. Says P.K. Gupta, Director, Intercontinental Operations, Legato Systems India and Chairman, SNIA India Chapter, “The essence of implementing ILM is to give greater importance to a data set that has the maximum business value, which more often than not is the latest data. Once we have classified data based on its value to the business, we can protect and manage it in a better way and do all that faster and with less resources.”

The trend toward virtualisation supports the emergence of technologies such as blade servers, on-demand computing, and n-tier server architectures. Says Sanjay Karade, Solutions Architect, Financial Services Industry, Cisco Systems, India & SAARC, “Virtualisation allocates physical devices into logical resources according to business need, allowing management flexibility to align storage, compute, and network infrastructure with an application‘s needs.”

But for ILM to work in the new data centre environment, you have to consolidate storage. This can be done by using either virtualisation or by putting a switch in the SAN environment. Within the new data centre, server capacity and storage resources are virtualised, data moves from one storage resource to the next based on the company’s ILM policies and storage provisioning tools dynamically reallocate storage. Servers and storage become inseparable as these technologies take root, the result is far more flexible, efficient and cost-effective storage environments.

• Network Infrastructure: Bank on it - It’s the backbone of every data centre, without which it is practically impossible to run any kind of system. Ideally it should be prepared to provide high-quality support for upcoming converged applications such as VoIP (Voice over IP) and SIP (Session Initiation Protocol). It could even involve the upgradation of core backbone switches to 10G Ethernet and the client systems to at least 1 Gigabit. Karade feels that as the data centre incorporates a large number of core switches, interconnectivity between these switches necessitates large amounts of bandwidth and 10G Ethernet is the optimal choice. “We have several customers deploying 10G Ethernet connectivity in their data centre. We also see Gigabit Ethernet deployed at the desktop level in many enterprises as access switches and network cards for PCs are available with 10/100/1000BaseT options,” he points. On the other hand Chatterjee feels that 10G Ethernet is especially effective in a disaster recovery environment where very high speeds are required to transfer a huge quantum of data from a backup server to the primary server or vice versa. “10G Ethernet is being seen as a very cost-effective way to upgrade data links in data centres as most servers being manufactured now come with gigabit connections as standard,” he says.
• Operating System: The war has just begun - Picking the best operating system from standards based Linux and Windows or proprietary ones such as AIX, UX and Solaris is a tricky matter. Windows requires other core Microsoft software components to support collaboration and integration of apps. Linux is growing rapidly and it well accepted in the server world. Proprietary OSs are used in scientific and mission critical data centres. One needs to have the best combination. For example, an ERP application can be piloted on HP’s IA64 line with Linux or Windows. You’ll also have to think of a platform for a new generation of Web applications. If you aren’t neck deep in standards, security and manageability issues surrounding these new applications, you will soon be if the choice of platform is not correct.
• Security: Hackers are on the prowl - Strong security is a given in any kind of new data centre environment, but with applications and access going global, threats loom large. In the years ahead, one will have to choose how best to deploy security throughout the new data centre. For comprehensive protection, enterprises have to manage not just security data generated by security devices like firewalls and IDS (Intrusion Detection Systems), but also data generated by network events and changes in configuration status of data centre servers, network devices, storage devices, and applications.

Today, enterprises deploy network management software to monitor network events; change management software to log changes and check those against how things are supposed to be configured; and security event management (SEM/SIM) products to help filter and make more sense of security events generated by firewalls, IDS and other security devices. But these solutions typically operate in isolated silos making it difficult to aggregate and transform raw data into actionable information. Says Capt. Felix Mohan, CEO Secure Synergy India, “Enterprises have to collate both security and management information through a single process, and centralise the information on an integrated management console. Event correlation technologies that are becoming common now would enable the console to make intelligent decisions and take proactive action to enforce security and compliance policies.”

Capt. Mohan further adds that the strategy would be effective only if all data access is conducted through a controlled gateway. However, a characteristic of today’s enterprises is unfettered connectivity that permits insiders to bypass centralised security controls, for instance through modems or wireless access points. “Focus on endpoint security controls is also vital. The technology controls should be supplemented with ongoing enterprise-wide security programmes to usher compliance with enterprise security policies and to protect insiders from becoming gullible conduits for malicious outsiders through social engineering attacks,” adds Mohan.

The essence of implementing ILM is to classify data based on the value to the business, protect and manage it in a better way and improve process speeds with lesser resources, says P K Gupta

Enterprises should ensure that their patch management technologies not only automate the patching process but also permit patch rollback and work in a heterogeneous environment (Windows, Linux and Unix). “In the new data centre, next generation patch management technologies will be required that perform regular vulnerability and compliance scans to locate systems where patches are needed, manage configuration policies, and permit testing of the patches in a simulated environment before applying them on production systems,” says Mohan. Says Karade, “The integrated security service in switches or routers should provide virtualised, hardware-based firewall capabilities between virtual LANs (VLANs), intrusion detection, Secure Sockets Layer (SSL) services, and VPN termination. Rounding out the security portfolio is an extensive list of software security features on routers and end-point host protection solutions.

Chatterjee feels that traditional security tools such as anti-virus software, password based authentication and firewalls are being strengthened with new tools such as intrusion detection and prevention systems, VPNs, content filtering tools, security auditing tools like penetration testing tools etc. “Enterprises are taking that extra step to ensure that their mission critical data is protected from worms and viruses, hackers and the competition,” he says. Even Managed Security Service Providers (MSSP) have the skill sets to audit, offer consulting or training, implementation and managed security solutions to enterprises. An MSSP can offer enterprises a cost advantage and bring in its experience to identify and ward off potential security threats.

l Network Management: It all ends here - It’s ridiculous to equate device management with network management. In fact the NMS should be clear cut in measuring application performance, resource management and allocation and offer appropriate tools to guarantee maximum application performance. The integration of network and security management is also a vital link in the new data centre. IBM Tivoli, HP OpenView, CA Unicenter TNG and Microsoft management tools - all are changing to gain impetus in the utility computing space.

In the end…

To harness the true value of the Internet and Web based applications, it is essential to gear up the data centre. All you need to do it stay focused on cost cutting while investing in products and services that move your company toward the new data centre. While doing so, you have to evaluate your key suppliers’ solutions to determine what actually works for you.

rahul@expresscomputeronline.com