Google Analytics

Search This Blog

Tuesday, December 24, 2013

Where milliseconds matter: low latency networks going mainstream

By Henry Bohannon, Senior Director, Head of Ethernet Product Management, Tata Communications

While the standard communications network is designed to offer a competitive cost structure, low latency networks have turned this model upside down: speed takes priority over cost.

Financial firms used to be the sole users of low latency networks. Financial firms generate revenue from the purpose built, low latency networks through arbitrage created between markets with electronic trading. Low latency links have been designed to provide the fastest possible speeds between electronic trading locations, allowing execution of trades between exchanges in milliseconds. Financial firms take advantage of the difference in the buy and sell price on different exchanges. The difference in the buy and sell price is extremely small, but by trading in large volumes, the firms can earn millions of dollars every day.

While firms in the banking, financial services and insurance industry segments are the primary users of low latency applications, other companies are now ready to pay a premium for the improved performance. This improved performance, between specific locations, delivers benefits that outweigh the higher cost. Low latency networks are of interest to pharmaceutical companies, IT organisations, cloud service providers and content delivery specialists. These are companies which are running business-critical applications or revenue critical applications which will not work optimally on networks with high latencies. It is no longer a simple binary question of an application working or not, but an investment in the network latency performance can deliver direct business benefits through improved application performance.



Several service providers offer low latency networks, providing options to the market. As the market expands beyond the financial services industry, new service providers are expected to enter the market with new choices. As the number of choices from service providers expands, choosing the right provider will become more complicated and time consuming. Here are some questions to consider:

Global, regional or metro low latency?

Customers need to focus on providers that can deliver their requirements - for metro needs, go to the local providers in that market. For regional low latency connectivity, the service provider options will change since not all metro providers go beyond the metro. Finally, global providers will be a separate set of service providers who have the expertise and investments on a global scale. For example, Tata Communications is a global provider, a regional provider, and within India, a metro provider. Phase 1 of Tata Communications' global low latency network links six of the most important financial centres of the world – Chicago, Tokyo, Singapore, Hong Kong, London and New York.

Point-to-point, or multipoint?

Traditionally, companies have worked with point-to-point links to build and manage the network they need. However, more and more companies are considering the cost benefits offered by multipoint Ethernet. Which solution do you need? The answer is very specific to your operations. Point-to-point links are common place, but ignoring the advantages of multipoint could put you at a competitive disadvantage.

Exchange to exchange, or city to city?

Traditionally, low latency networks were built extending directly into the data centres where electronic trading is taking place to achieve the highest possible speeds for financial transactions. While this solution fits well with financial firms, the electronic trading centres may not fit the needs of other industries.

What is being monitored?

Customers need to be cognisant that the overall circuit performance is very important. Continuous measuring of latency is critical, but packet loss and availability are also factors that need to be monitored. The service level agreement (SLA) proves how confident the provider is in their network. Companies with good SLAs monitor the latency of their network every second, while online reporting refreshed every minute allows customers to visualise link utilisation, packet loss, availability and jitter to ensure the link is operating at top performance.

Do you need diversity?

What is your business model? How much diversity or availability do you need? Do you only need one unprotected link with the absolute lowest latency or do you need low latency with high availability? In some regions, the lowest latency may be on a route that has frequent outages. Low latency networks are often sold as unprotected; the customer can purchase multiple links with the best latency and perform their own protection switching keeping direct control of the latency performance. The customer business model will determine the risk and the amount of effort that needs to be invested for building a network that achieves the best results.

Do you only have low latency needs?

Some network providers specialise only in low latency networks, whereas others may also offer standard networks that are not purpose built. If you do not require low latency to every location, then working with a provider that operates multiple networks will save you money. This is true, especially for multipoint services. Providers who operate multiple networks can deliver a single multipoint solution over multiple networks, delivering the best solution for the best price.

Other factors to consider

Low latency networks do differ in cost and speed. Is the extra speed worth the extra cost? What about security? Will your data be secure on the service provider’s network? The financial strength of the service provider is also important, especially if you are running a significant part of your business on the network being provided. No one can afford to wake up one day and find that their service provider has gone out of business and their network is permanently down.

The low latency market is expanding beyond the traditional financial institution user. On the surface, low latency networks appear to be all about one thing, latency. In reality, new and old low latency users need to ask a number of questions to identify what they need, the risk they can afford and what the best solution is for their requirement. Networks are a vital part of a company and often, the revenue generators for the company. A company with the best network, the best price and with acceptable risk will successfully maximise revenues and efficiencies.

No comments: