Building differentiation in the datacenter market

In this new series on how datacenters in India are building differentiation in the market space, we speak to P Sridhar Reddy, Chairman & Managing Director , CtrlS Datacenters

At last count there were over 70 outsourced datacenters in India. Does that mean that the outsourced datacenter market is becoming commoditized with all the competition?

 

That number is not a real representation of the market. Of these 70, many are not commercial collocation facilities. They are either captives or are run for specific clients only. And many others have reached a level of utilization where they cannot take on newer tenants. Like our Hyderabad centre, for example, they many still rent out some small space, but they are no longer in the market. And some players are in the early stages of consolidating their physical locations.

 

At this point in time, there are no more than 10-15 datacenters allover India, that are seriously in the market for serious hosting deals. Now, if we get region specific, the numbers are even smaller. Take Mumbai, the biggest market. There are only three contenders here who compete in the collocation business. There are a few small facilities that will be in contention for the smaller, one rack – two rack deals, but no one else in the collocation space. Similarly in Delhi there are three.

 

Are you saying that there are not enough co-location players in the country? 

 

No, all these put together, we have about a million square feet available. So, there is enough capacity available for current needs and for the next five years. Basically, CIOs can pick and choose what they want. In managed services, it is a more limited market.

 

How do datacenters differentiate themselves from one another?

 

The market can be divided into 3 or 4 segments. There is collocation; there is managed services; and there is cloud, hosting and so on. Like I said, the collocation business is a pick and choose market, so some players tend to play on price. In managed services on the other hand, players do not get into price discounting. 

There are many other ways a datacenter can differentiate and position themselves in the market. Let us take some US examples: Softlayer has positioned themselves as the best bare metal server player in the world. In services, Rackspace beats them. Rackspace was positioned for the hosters and SMBs of the world. I have not tracked them for a while, but back then, Savvis was positioned for the enterprises of the world. Obviously, they were very expensive, but came come with many more processes. 

 

Equinix has mastered the carrier business and Digital Reality mastered the real estate play in this business, and so on.

 

Coming to the Indian market, what is your approach to this market?

 

Our approach to the market is very simple. Any market segment we enter, we want to become leaders. Let us take collocation for example. Now, how does this come about? The key is our five nines SLA (99.999% uptime). For a large bank even one minute of downtime is millions of dollars of losses. Remember that a minute of datacenter downtime converts to more than an hour or two of application downtime, when you count all the reboots and resets. We have done over 4 renovations of our datacenter, without any downtime.

 

Then we have done about 70-80 innovations in datacenters. For example, we have won a national award for innovation datacenter air conditioning systems. Seven years back, I was going around the datacenter when I saw that the there was hardly a few racks in one area. The air conditioners were on and all the fans were working. The fans consume about 30% of the power used by the air conditioning systems. I said, why should all fans be on? Can’t they be intelligent? None of the vendors supported us. We did the R&D to develop intelligent fans for datacenter air conditioning. That won us a national award. Today all datacenter air conditioners come with intelligent fans. We have innovated on some of the basic design elements of a datacenter. Traditional datacenter design was based around the watts / sq ft loading expected. We changed that concept.

 

We are now helping high end users like the NSE and ONGC to design their datacenters. That looks like becoming a new business line.  

We have innovated in business models also. DR on demand is an example of this. One of our customers said that DR costs were too much for him. Can I share that cost with someone, he asked? That was the Eureka moment that lead to our creating the DR on demand option. 

But our core proposition is our SLA. If we don’t deliver on our SLA of five nines, don’t pay us. And we have not had a single outage in our eight years of operation.

 

And how do you ensure that the SLA works?

 

First of all there is the physical infrastructure. Everything is duplicated. We have two sub stations. You connect then to two different power producers. Then we have two DG banks. Each one is capable of supplying uninterrupted power to the full datacenter.

 

We have fuel reserves for 4 days on full load. Normal load is under 50 %, So, that is 8 days. The fuel tank itself is a big project costing a few crores! Our Mumbai datacenter fuel tanks are bigger than that of any gas station. Each has dual power lines and pumps, automatic level detectors and intelligent controls. We store water for 10 days. That is how it builds up.

 

The design is done in such a way that we need not ask clients for planned down time for datacenter renovation for 30-40 years. And in these 30-40 years, all datacenter equipment will be replaced. Some, may be twice or thrice over. But the datacenter will run through all these without disruptions.

In Mumbai, we have chosen the site such that there are two sources of gas nearby and we can build a 16 MW captive gas based power plant if required.

 

How are your teams geared for fault tolerant operations?

 

We assume that human faults will occur. The design is in such a way that even if those errors happen, they will not bring the systems down.

For example, the worst thing that can happen has happened in our datacenters. Two power sources got mixed. This should have lead to downtime for days, but nothing happened. It got cut off. That is the way we have designed our safety systems.

 

Somebody could accidentally switch off something, a chiller plant or a UPS for example. If a chiller plant for example goes down, the other one picks up. If a valve goes back, there is another route. 

 

What is your differentiation in the cloud space?

 

The cloud space is interesting, but we have not fully conquered it yet. The big players are going to be always ahead of us when it comes to Orchestration layer capabilities. This is a must for the US market, which is a do it yourselves market. Do it yourselves is not very popular in India. So we are not going to get into that play.

 

We can be better than the big players on cost comparison. What we can’t do is match their brand pull. But I don’t think that brand is as important in the B2B space as is quality of service. So, the unique value proposition we built is the tier four cloud or the four copy cloud. It is a sweet spot for mid to large enterprises. It automatically makes four copies of your data to ensure data availability. Two copies are in the primary datacenter. The fourth copy, in a different datacenter works like a tape backup to protect against data corruption. 

 

We are going international with this offering. We are going to 17 countries in the Middle East and Africa. We already have some significant wins there.

 

All these extra measures will come at an extra cost for the user, right?

 

We are lucky that our competition is not efficient. So we are able to manage growth rates and profits in the 40 % range and keep ourselves financially healthy. This is a very capital intensive business. It can get so hungry that cash flows could get completely dried up. So, I have kept it very very disciplined. Next year our EBITDA will be equal to our entire debt. Our original plan was to build four datacenters at one go. But we have deliberately gone slow. What I mean is that we have chosen slow growth with a quality product.


Add new comment