Need to define outcomes before making data centre software defined

There is a need to clearly define the outcomes from adopting the software defined concept; a necessary mechanism to regularly review and revise the outcomes should also be built, says Manoj Gautam, VP - IT Infrastructure Service, Maruti Suzuki India

In the new age IT, mobile, cloud era, the expectations of the functional users from IT is changing rapidly. The desire is to get services with efficiency, agility and security. In this backdrop it’s necessary to take a step by step approach to build gradually. The IT should have two prerequisites – to become an enabler for business and it should become a service broker for both the private and public cloud services by establishing a unified operations environment and framework.

Software defined IT infrastructure can serve the purpose of being an enabler and easing resource provisioning and management. Maruti Suzuki began its software defined journey long back with server virtualisation . “We shall be implementing software defined networking solutions in the near future. Storage isn’t on the cards right now but will be taken up later,” informs Manoj Gautam, VP – IT Infrastructure Services, Maruti Suzuki India.

Advice to peers on software defined data centre
Gautam’s advice to companies exploring the software defined elements for their data centres is they should clearly define the outcomes from adopting the software defined concept; a necessary mechanism to regularly review and revise the outcomes should also be built. “The major outcomes for us are: hardware provisioning should become easy. Additionally, the environment should be trouble-free to manage and monitor. Secondly, it should result in increased automation. For example, in case, the threshold is breached, the automated alerts should be relayed to the concerned people. Service delivery should also become more responsive on account of ease in configuring and reconfiguring the systems. Thus the security and policies can be managed under the ever changing software defined workload environment. For us, reducing the manpower doing the mundane monitoring jobs is also one of the outcomes. We are expecting a reduction in efforts to an extent of 10-15 per cent. The time thus saved may be spent on other value adding jobs,” says Gautam.

Hardware is equally important in making the data centre software defined. It’s a myth that hardware doesn’t matter much while adopting a software defined environment. The hardware derives the importance from the fact that additional tools that run on the server hardware should be compatible. The concerned staff in the data centre should be trained extensively in the software defined domain. If necessary, to get them certified from the OEMs. “We cannot just rely on the staff from the OEM and training our manpower is equally important,” feels Gautam. Lastly, security should be not ignored at any cost.

An issue faced by enterprise running captive data centres is the lack of talent having a comprehensive understanding of all the components. There are super specialised people with niche understanding. Gautam has solved this issue by having certified a few personnel in the cloud domain and the overall area of data centre management. “We have a few personnel who have gone through rigorous training in cloud and the overall data centre management. They have bagged the relevant certifications,” informs Gautam.

The captive data centre has over 60 racks with a mix of normal and blade servers. It also has High Performance Computing (HPC) servers suitably enabled with in-row cooling. The DCIM and other monitoring solutions are
in place.

Steps for energy management
Server consolidation is one of the first action points for energy management. The cooling should be efficient and the cold air wastage is tracked and in case there are leakages, they are plugged immediately. “We have gone for in-row cooling for the HPC servers. The cooling tower sits between two racks and circulates the air, which doesn’t go out. It’s a kind of geofencing of the air. So, it doesn’t get wasted,” says Gautam. The DCIM system keeps track of the important metrics to be tracked for power consumption.

Maruti Suzuki regularly invites IT infrastructure environment experts for checking the vitals of the data centre. “They would probe the quality of power, the presence of suspended particles in air, water and fire systems, etc. The data centre audit is done regularly,” informs Gautam. UPS is an important component in energy management. The company has a monitoring system to keep tabs on the UPS in terms of the battery quality, sectors of the UPS getting weaker, charging, etc. As far as cloud is concerned, a hybrid strategy has been adopted, “We have adopted private, public cloud for various applications,” points out Gautam.

Data centre security
The best of breed security solutions have been adopted to secure the data centre. “The Network Access Control (NAC) assures no external device can be plugged into our enterprise network – in addition to the Security Operations Centre (SOC) we also use spam filtering, data leak prevention, etc. The network segmentation feature is also being planned in the near future. Specific to DDoS, there is a dedicated solution to guard against attacks,” he mentions.

DC availability is always at the helm of all the require-ments. The company, twice a year, does DR drill for various applications from the DR site in order to check the application robustness.

The future Caution will be exercised in the cloud journey. Only the non-critical applications are planned to be put on the public cloud. “We are actively exploring emerging technologies like the Kubernetes based containerised application management. Solutions are also being explored to have complete visibility of the data centre – compute, network, storage, UPS, electricity, data centre environment, etc,” concludes Gautam.

Manoj GautamMaruti Suzuki India
Comments (0)
Add Comment