Issue 8, Feb/Mar 10
FOCUS FINANCIAL MARKETS
demands and build additional capacity.”
Three of the fi ve 20,000 sq ft modules of the Mahwah facility will open on day one, with the other two left for future expansion. The 400,000 sq ft building’s power capacity will be 28MW. Power densities on the raised fl oors of its data halls will go up to 150 watts per square foot.
NYSE adopted a modular design for the Basildon data center as well. It will house a total of seven 3MW pods, each providing 10,000 sq ft of raised fl oor. This is also a 400,000 sq ft building, sitting on a 16-acre property.
Besides the constantly growing complexity of client algorithms, a major demand driver for NYSE data center capacity is the operator’s allowing member companies to provide “sponsored” market access to their own clients – the only way for non-members to use the exchange’s colocation space. As members grow the number of clients they serve, so does the amount of space they use in the data center.
LONDON SNOWBALLS LSE began its colocation business with a small pilot in September 2008.
As a data center owner/operator, LSE had experience of meeting user demands but, according to Nigel Harold, head of business development technology for its hosting exchange service, the development of the colocation business meant adapting procedures and policies to accommodate new tenants that wanted fast connection to its Tradelect platform.
“We had a number of requests from proprietary traders and worked with them to determine what they were after. We made a small area available in September 2008 and it kind of snowballed,” says Harold.
The model changed as the exchange quickly saw that connectivity was key to effi cient operation of trading algorithms and it expanded its carrier footprint. As proprietary traders, the clients choose their own carrier and develop their own algorithms.
“We help with installation – racking and stacking – and ongoing management. We offer remote hands and refi ne what they need,” says Harold.
Because it is based in central London, the exchange charges on power consumed with two levels – 3KW per cabinet or 5KW per cabinet – on offer. Clients are trying to squeeze as much capacity as possible into
their cabinets and various cooling methods are employed as “clients are quite power- intensive”. The site is N+2 and all cabinets are monitored to ensure clients stay within their limits.
LSE won’t say how many clients it is housing in its data center, but does say that demand is such that it wants to release further capacity and is refurbishing other fl oors.
Recent developments at the LSE include the opening up of a new fl oor to house members’ equipment and a deal with market data aggregator Quanthouse.
The exchange is even considering migrating its own equipment out of its data center in order to free up more capacity for traders.
On the deal with Quanthouse, the market data on the number of asset classes being sent meant it made sense for them to take the data internally from the exchange, says Harold.
In September 2009, it bought MillenniumIT, which had its own in-house developed trading systems with multi asset-class functionality.
It looks as if the exchange will upgrade to a 10Gb Ethernet networking infrastructure, but Harold has not ruled out a move to Infi niband – the ultra-fast networking architecture.
Along with the capacity, the cross connects and the low latency data aggregation, the exchange is exploring whether it can provide order-validation services. With new exchanges (see article next page) and competitors’ data centers coming to the marketplace, there is no choice.
DATA FEEDS Beyond the trading platforms, the data providers are trying to cut latency by providing hosting in their data centers.
Thomson Reuters (TR) provides colocation and managed services for traders in its data centers, along with its market data feeds. “We have headroom of way north of 20%,” says Bill Ruvo, global business manager for the company’s Real-Time Data Feeds division.
“Given that we have been in the market-data infrastructure business for years and years, we have tight processes for capacity management.”
In October, TR announced a new partnership with Savvis, a large US-based data center provider that will host infrastructure that supports TR’s new managed hosting solution, as well as colocation space for fi rms that wish to install their servers next to that
"They're already building in a level of growth" – Young infrastructure and connect to it.
The deal includes Savvis data centers in New York, Chicago, London, Frankfurt, Tokyo and Singapore. The company planned to launch the offering in all regions except Asia by the end of January. Tokyo and Singapore were due to come online mid-February.
Prior to the announcement of its deal with Savvis, TR has been using two BT Radianz data centers in London and New Jersey. The company partnered with Savvis to expand its global reach in proximity to major trading markets. “We’ll continue to service the customers that are running in those BT data centers until the end of their contracts,” Ruvo says.
The company does not outsource data centers that support production of its core product – consolidated data feeds from hundreds of exchanges around the world – keeping that task in-house. “Thomson Reuters data centers are used to produce and distribute our standard products to customers,” Ruvo says.
VERIZON BUILDS TO STAY AHEAD Verizon Business, which runs the data center infrastructure that hosts the IT equipment of Nasdaq OMX, is also making expandability a key part of its offering to the exchange operator.
“The reality is at some point in time your data center infrastructure will hit a limit,” says Verizon’s executive director of Technical Operations, Brian Trosper. “Can you expand (in an existing facility) or build in a nearby or adjacent space?
“We are expanding now to stay ahead of demand. We purchased enough real estate – raw land – so that we can handle expansion of data centers within the same geography,
Continued on page 28... www.datacenterdynamics.com
| Page 2
| Page 3
| Page 4
| Page 5
| Page 6
| Page 7
| Page 8
| Page 9
| Page 10
| Page 11
| Page 12
| Page 13
| Page 14
| Page 15
| Page 16
| Page 17
| Page 18
| Page 19
| Page 20
| Page 21
| Page 22
| Page 23
| Page 24
| Page 25
| Page 26
| Page 27
| Page 28
| Page 29
| Page 30
| Page 31
| Page 32
| Page 33
| Page 34
| Page 35
| Page 36
| Page 37
| Page 38
| Page 39
| Page 40
| Page 41
| Page 42
| Page 43
| Page 44
| Page 45
| Page 46
| Page 47
| Page 48
| Page 49
| Page 50
| Page 51
| Page 52