HK Jockey Club: Aggressive yet cautious on cloud

By Carol Ko 07-Jun-2010

http://www.asiacloudforum.com/content/hk-jockey-club-aggressive-yet-cautious-cloud?utm_source=lyris&utm_medium=newsletter&utm_campaign=daily

Virtualizing the IT infrastructure of The Hong Kong Jockey Club (HKJC) is no mean feat. Operating by seven divisions and 51 departments, the HKJC currently employs 4,600 full-time and 20,300 part-time staff. Doubtlessly, the HKJC is one of the world's largest horse racing organizations.

Cloud computing, at least at the private cloud level, is being aggressively pursued at the HKJC. To date, the Club has virtualized 10% of hundreds of its servers, mostly in its testing environment. Moving on, the Club will cautiously test out cloud computing capabilities in the development and production environment.

In this interview, the HKJC's Executive Director of IT, Sunny Lee (picture), shares his experience of his firm’s cloud computing adoption at the Club and what the major challenges are. As former president of the Hong Kong Computer Society (2007-2010), Lee also commented on the prospect of turning Hong Kong into a cloud computing hub in Asia.

"Our biggest challenge is really staff, talent and mindset—things beyond technology." -- Sunny Lee, executive director of IT, The Hong Kong Jockey Club

Asia Cloud Forum: Will the Hong Kong Jockey Club go 100% into the cloud?

Sunny Lee: As with other organizations, some of the applications may not be appropriate for the cloud. We would definitely not go 100% into the cloud because there could be areas where we require high security or better control over the performance—or the fact that there might be no economies of scale to begin with from the cloud. 

It doesn't mean the cloud is not good. It's just that there's no one-size-fits-all. We'll be pursuing cloud technology, and we'll be taking it one step at a time. 

What HKJC applications will you take to the cloud?

Lee: What we're doing today is really at the beginning stage, though we've done quite a bit. From the scale's perspective, we've pretty much competed the virtualization of our testing area. It's now ready for a lot of sharing from the same platform and the same hardware. And we're already reaping a lot of benefits.

In terms of scale, we're talking about [virtualizing] hundreds of machines at this point in time. But there's only a small fraction of everything that we have virtualized—about 10% of all our servers.

Our next step is to start virtualizing the production environment, beginning with some of those corporate-related production infrastructure. At the same time, we'll be virtualizing the development environment. 

We're quite aggressive and cautious at the same time. Most likely, we'll be more interested in building a private cloud rather than a public cloud. We have quite a bit of scale to begin with, as we have a good population of machines to work from. But so far, what we've been doing is still largely confined to server virtualization, and the server cloud type of technology. 

We have not yet ventured into, say, database, storage or networks. And we don't see that to be of the highest priority. We want to begin with server virtualization, reap the benefits, and learn from it. 

With the result of virtualizing the Club's IT assets, what have been the biggest business and operational benefits you've achieved so far?

Lee: What we've seen so far might not be comprehensive and we'll be able to see more is the fact that we can actually get our machines to higher levels of utilization as advertised in the market. And because of that, we're able to save a lot of capex and opex related to those machines.

The whole system from the management side is becoming simpler, as we're virtualizing and standardizing the whole technology stack. Because of that, we are able to gain efficiency in terms of system deployment, while compatibility issues will become less of a big issue.

All these things are not too far off from what the technology providers have been advertising about. We're quite happy with that. And certainly we've learned quite a bit along the way.

The HKJC has a huge IT team. How is the Club's IT division coping with re-skilling its IT staff from utility management to application development? 

Lee: Our biggest challenge is really staff, talent and mindset—things beyond technology. Of course, the stability and maturity of cloud technologies are important for us. But of equal importance is the readiness for the organization—the people in particular—to embrace the concept of cloud computing. 

In the past, everybody wanted dedicated machine for their applications. As when performance issues arose, at least there was one less thing to blame on. But then it's inefficient. 

It's difficult for the different business units to change their whole mindset—to be willing to share the same machine to run their different applications on—especially with different applications owned by the different business units. Since the business owns the budget, they may ask “Why do I have to share this server with other people?” 

At the same time, we have to stress on the whole concept of service-based computing—meaning that we'll offer IT as a service, rather than a piece of hardware or software—something you may not be able to touch and feel, but will be able to satisfy your requirements.

What about security concerns around cloud computing? 

Lee: Security takes a different level of attention when you start virtualizing applications and sharing apps on the same platform. We have to be more capable in terms of security management in a virtualized environment. And that's why we started with the non-production environment (that is, testing and development environment). So that we can have a chance to really experience the real issues involved, understand the intricacies and complexities and the specific issues related to the virtualized space.

Even if we roll out cloud computing in the testing and development environments, it has already given us a lot of efficiency gains, as we're able to buy fewer testing machines. 

The staff mindset needs to be changed. It will take a lot of enhancement skills and technical capabilities—but that's the easy part. Once the mindset issue is tackled with, it is relatively easy for the person to get changed from the technical and operational perspectives. 

Cloud computing is not insurmountable. But there's a fair amount to learn about it. That's why I do suggest that if you go in the cloud direction, you should really take one step at a time. Carve out a small pilot, involve a good miniature version of what you're going to do and prove it—never the mission-critical ones though. This way you can really get your people and your organization up to speed.

Other than the HKJC, what other good examples of cloud computing adoption have you seen? What makes them successful?

Lee: I think this is still a starting point, I've seen a lot of good public cloud type examples of which I'm personally very excited about. Google and Microsoft offer good examples for SMEs to pick applications which can help automate their businesses.

As for public cloud adoption at the HKJC—I think it's still too early to say, though I'm confident with the success demonstrated by the public cloud, this Software-as-a-Service offering, I'm confident that the big enterprise will be able to see some good successful implementation coming up shortly in the next 12 month or so.

When I'm wearing a community hat [as former president of the Hong Kong Computer Society], I think Hong Kong is in the best position to run a cloud center or hub. 

But does Hong Kong face a data center crunch?

Lee: The data center crunch is certainly a big issue. Everybody I talk to, they tell me they're already running out of [data center] space, and we've already given up some good data center business to other countries. But I'm encouraged to see that the Hong Kong government is getting more proactive. Hopefully there will be more promising discussions and considerations. 

What I'd like to see is a lot more focus on building Hong Kong as data center and cloud hub in the region. Leveraging our good infrastructure to the fullest extent, in terms of electricity, our strategic geographical location, low telecommunications cost, the proximity to China, good rule of law, free economy, duty-free port and the availability of talents, there are a lot of good reasons for us to sort out this facility issue—the availability of land. 

I'd really like to see a critical mass created perhaps like a park allocated for data centers, where all the good infrastructure will be pre-provisioned for data center space, thereby not requiring operators or users to go through this whole lead time and all the associated set up work related to getting the last mile on telecommunications and power.