ONGC is the most profitable PSU and the business involves a heavy technology dependency. The data centre is the heart of keeping the IT systems available at all times.
The data centre transformation story at ONGC began in 2012 with server virtualisation. It provided with the necessary operating flexibility, reduction in the number of machines and thus in cost reduction. “We have one of the largest SAP application suite running in the data centre. VMware provided the virtualisation layer and HPE is our hardware partner,” informs Dharam Raj, CIO, ONGC.
Virtualisation – a step towards becoming software defined
ONGC has also virtualised the network. Moreover, the company invited one of the largest tenders in the storage space (over 500TB for the ERP data centre), taking the flexible capacity model i.e pay-per-use. “This is the largest such contract of its type for a captive data centre in India. We have partnered with HPE for a multi-year contract. This solution (with all flash storage) will have a software defined environment. It will consolidate the storage units across data centres and will also take in data residing in the operational technologies like SCADA,” says Raj. Usually, the data residing in the OT systems in the companies remain isolated and it’s only used for accounting purposes. The need is to extract business context from the OT systems.
The all flash storage is AI friendly, “The advantage with all flash storage is it makes analytics, AI and ML operations very smooth. The disk and other storage systems will not be able to deliver the same performance when cleaning the data and also running AI, ML on it,” states Raj.
ONGC runs a dedicated data centre for ERP applications (150 servers). Similarly, there are separate data centres for scientific applications and email, other applications. The current focus is to consolidate the business applications into one data centre to save costs. “It will result in cost savings of up to 50 per cent – due to optimisation in cooling costs, hardware rationalisation, administrative and other costs,” informs Raj.
Data centre security
When asked about information security arrangements made in the data centres operated at ONGC, Raj stressed on the importance of the granular understanding of every component of the data centre. “We regularly conduct training on the latest in information security for the staff. An Integrated Security Operations Centre (ISOC) is also operational. Most of the breaches happen because of the hygiene measures not followed and misconfigurations. Thus the triad of Governance, Risk and Certification (GRC) is crucial with proper change management practices followed,” says Raj.
In case if an application administrator is making a change, then the server and network administrator also needs to be informed. The change document has to be routed with the concerned heads before making the changes.
ONGC was amongst the first PSU to be certified with ISO 27001 certification. However, certification for the sake of it will not be the saviour. A thorough understanding of the implication and the meaning of each and every guideline will go a long way in keeping the data centre environment secured. “We got the ISO 27001 certification way back in 2007 and we are working towards getting ISO 38500 certification. The necessary steps are being taken to attain it,” mentions Raj.
Taming pollution
One major challenge of operating data centres in the National Capital Region (NCR) is the heightened level of pollution affecting the data centres resulting in IT hardware corrosion. In fact, there are cases of companies having to move out their data centres of the NCR region or move to a hosted model because of the pollution surge. “We have deployed advanced filter technology to tame the adverse effects of pollution. It’s a three layer filter technology, wherein two layers are continuously working and one filter is on standby. The pollution is measured in V1 and V2. It has to be at the V1 level and the reading is taken in every shift. We haven’t had a single instance of planned or unplanned shutdown in the data centre due to pollution, since implementing the filter technology,” informs Raj.
He has been successfully able to provide 100 per cent data centre availability for many years at a stretch.
Best practices for running a data centre
For any data centre to be up and running, there are many guidelines to be followed. According to Raj, the first and foremost is a complete understanding of every component of the data centre. Secondly, it has been found that data centre downtime doesn’t usually occur because of a server or technology related failures but because of electricity shut down, pollution related failures, etc. Thirdly, the DR systems should be tested on a regular basis. All the applications should be tested for smooth functioning during disasters. It has been found that applications do not boot up during disasters because they haven’t been tested for the environment. The teams should write the testing reports every time the testing is done and then a follow up should be done on the errors being rectified from the previous testing sessions.
Change management is another important aspect – the relevant stakeholders should be taken into confidence before effecting a change. “I don’t allow any electricity shutdown without getting the consent of all the teams,” says Raj.
Another critical component is taking care of the UPS. It should be mandatory to keep the UPS battery room temperature at 25 degree celsius and the shift people should make sure this happens. The air condition should also be operated in tandem. There are over 25-30 critical points to be taken care of by the operations team and they should stick to the adherence. Any exception should be reported immediately without any delay – be it any time of the day.
As a regular practice, Raj takes the signature of the concerned person who is approving of a change to be made. “Many a times, we see people saying that they weren’t aware about a certain change being made in the system. Instead of mail communication, I still prefer a signature because it brings more accountability,” points out Raj.
As a part of the education initiative, it’s important to “make a lot of noise” about the incidents / mishaps happening in the data centre. It brings awareness and makes sure the same mistakes are not repeated.
Future of data centres
Companies in India will start consolidating the multiple data centres located in different geographies. A similar trend can already be seen globally with Fortune 500 companies. Over time, companies will move to pay per use model, which eliminates the hassle of technology purchase. It has the potential to reduce costs by as much as 50 per cent. “An interesting aspect with the consumption model is an automatic technology refresh after five years without extra cost,” shares Raj. Sans the consumption model, it will not be possible to extract the best value from the new generation technologies like AI, ML, etc.
The flight to the public cloud will be gradual. In the captive data centre space, the data centre personnel are super specialised but there is lack of talent having knowledge about all the components and a comprehensive view of the data centre.