The push for digital transformation is creating new challenges for MSPs, especially those that are supporting clients looking to leverage the hypercloud.
“Latency tends to rear its ugly head when you least expect it,” said Raj Mehta, CEO of Infosys International. “Our customers do not want to experience latency issues when moving to a public cloud."
Therein lie the real problems with latency. Users exposed to the performance and throughput of applications running on a private or hybrid cloud, where those applications are hosted in the data center, have come to expect a certain level of performance. Moving those applications to the public cloud adds latency if access is not set up efficiently.
“Latency is increased when you add distance and hops between the application, the data and the user," explained Michael van den Assem, managing director at Interxion Netherlands. "When you think of typical connectivity to a data center, the user accesses the resources [housed] in the data center via a telecom provider; that means some routing and latency is involved.
“At Interxion, we reduce latency by working with the telecom providers to make sure each has a node in our data centers, reducing the number of hops needed to connect to resources," he added.
Interxion operates 50 data centers across 13 European cities, all with multiple fiber connections, providing global coverage.
“Interxion's connectivity to the Amsterdam Internet Exchange creates a low-latency network that allows network operators to securely colocate in immediate proximity to the exchange’s IT infrastructure,” van den Assem said.
Simply put, data centers that interconnect with large exchanges reduce latency and transit costs by peering directly across the exchange’s switch.
That proven ideology has reduced latency significantly for those using colocation services from providers such as Interxion; however, that's only one piece of the latency puzzle. Other factors come into play, such as connectivity to the colocation site that telcos and other connectivity providers offer.
“Here at Interxion, we take a carrier-neutral approach, which mitigates the unpredictability of latency and provides direct connections from our data centers into the carrier networks," van den Assem said.
“Having dedicated and fast connectivity to a colocation site definitely proves to reduce latency and takes some of the guesswork out of determining application performance," said Mehta. "However, once you add a public cloud to the mix, things become less predictable.”
Encountering latency when accessing the hypercloud proves to be a somewhat different situation. MSPs, colocation sites and data centers seem to have better control over latency when the connectivity support is focused on providing access to a customer's ISP; in other words, a simple connection, starting from the customer site traversing via a telco and arriving at the data center, can be more easily measured, controlled and addressed. Once an external element such as a public cloud is added to the mix, control of connectivity becomes somewhat murky.
“Connecting to the public cloud from a data-center presence usually means that additional connectivity via another telco is required, that in turn means adding more hops to the route, significantly slowing down response times," said van den Assem. "That manifests itself with increased latency.”
That increased latency can have a negative impact on hybrid-cloud scenarios, where applications might not deliver the performance desired; yet, there are ways to mitigate that issue, van den Assem says.
“Data-center and colocation operators must partner with the public-cloud providers to limit that latency, something that may be critical for line of business applications," said van den Assem. "Here at Interxion, we have agreements with Microsoft, Google, and Amazon to interconnect with their respective cloud services. Each of those providers has a physical presence in our data centers, eliminating the need to use external cross connections.” van den Assem said.
When it comes to combating latency and building out a hybrid-cloud strategy, van den Assem's sage advice brings forth the notion of looking at the underpinnings of any colocation provider to prevent a nasty surprise when it comes to latency. MSPs pushing into the hypercloud and hybrid-cloud markets should look for providers that have done as much as possible to limit the number of hops and also have partnered with telcos and cloud services providers to make sure that interconnections are accomplished locally at the colocation — and that both have a physical presence there as well. That will go along way to bringing latency under control, while also lowering connectivity costs and improving the privacy of connections.