Melissa Swartz | No Jitter | June 14, 2018
Advice from the trenches on moving communications systems to the
At Enterprise Connect in March, I presented a session about avoiding the “gotchas” in a cloud migration. There were five panelists who discussed the issues with me, and who had some great advice for the audience:
- Mehdi Salour, SVP of network operations and DevOps at 8×8
- Skip Chilcott, global head of product marketing at IR
- Curtis Peterson, SVP of cloud operations for RingCentral
- Jamshid Rezaei, CIO at Mitel
- Chad Elford, global director product management, business collaboration – UCaaS solutions at Tata Communications
Cloud’s Network Impact
In addressing the technical gotchas to avoid, most of the discussion centered on how a move to the cloud impacts an organization’s network. The demands on the network change, and the ability to see what is going on becomes more critical. A network assessment is an essential first step.
“When moving to the cloud, you are adding two additional segments to your network: the connection to the cloud provider (either a public Internet or private connection), and the network inside the provider,” said Skip Chilcott, of IR. “Having visibility into this is important. A network assessment is critical. A lot of organizations give a lot of consideration to the connection to the cloud, but they haven’t thought about what their own network looks like internally.”
Network assessments should be performed far in advance of a cloud migration, Chilcott said. “And don’t just do it once. Do it all the time,” he added. In fact, a common mistake IR sees its customers make is running the network assessment for an hour, looking at the results, and deciding they have great performance.
Another common error IR has observed is something called the pilot trap, Chilcott continued, which is where an enterprise runs a small pilot for a while and everything looks great, so they’ll disconnect the pilot, move it to the next testing environment and test it, and so on. But network assessment is a constant thing that you need to revisit on a regular schedule — it’s not something that’s “one and done,” he said. “A network assessment in the planning stage is critical, but never stop, because there are blips and issues that happen. You can have serious degradation, but unless you are watching for it, you’re not going to see it. It’s not just in the cloud, but between sites and different locations within sites.”
It’s important to have a strategy to deal with those blips and degradation, Chilcott concluded. “And you have to think about the end-to-end connection, all the way from the headset, the dongle connecting to the PC, through the PC, to the Wi-Fi, and through your network. You have to think about all parts of your network,” he said.
Talking App Availability and End-to-End
Chad Elford, from Tata Communications, added to Chilcott’s advice, emphasizing that it’s not just about the network, but also application availability and the routing required to get that real-time media and quality of service (QoS). Using network diagnostic tools is one thing, he said, but being able to know you’re on-net and that your executive will be able to get on a video conference call quickly, and with excellent quality, is another.
“Does your service provider offer a service-level agreement that is end to end, or ‘ear to ear,’ with accountability from the user to the cloud and to the other end?” Elford said.
“Otherwise, how do you make sure the network works in multi-clouds, end to end, regardless of location? You should only move service or application to the cloud if you’re sure of the connectivity to the cloud and back. I’ve never met a customer that didn’t do a production pilot. And you have to continue to re-assess on a regular basis.”
Picking the Right Tools
Infrastructure and the tools to test and assess are extremely important, agreed Mehdi Salour of 8×8. “Tools are something that you should pay attention to,” he said. Some providers offer a very simple Web GUI, others offer tools that include multiple components. Ask your service provider what tools they provide to do a proper assessment. It goes beyond just jitter and packet loss. How is your DNS reacting? Is your NAT broken? Is there packet fragmentation going on? Is your NTP working properly? There are so many things that need to be tested on your network, so it’s important to ask your service provider if [it] can provide you with the tools to help you facilitate these tests easily in your environment.”
In today’s world, Salour added, the fact is you’re going to have remote and mobile users, which means you’re going to have users on networks over which you have no control. Not all solutions are built the same way, he said, and it’s important that the underlying technology is capable of providing services in less than ideal or adverse network conditions. Buffer bloat, for example, is detrimental to voice quality, and it happens more than people think, he said.
“Include remote and mobile users (3G, 4G and LTE) in testing to ensure that your service provider can provide you with the best quality of service no matter where your users are,” Salour concluded.
When evaluating the tools you will use for monitoring, look for companies that provide tools both for instant and long-term visibility, advised Curtis Peterson of RingCentral. “That cable service that you are using for half your employees might be pretty good, but then it might start to degrade. You need to be able to diagnose that. You need to be able to look at a call and determine that, ‘Aha, these guys are part of my problem.’ You need to be able to diagnose any network at any time.”
Enterprises should look at a vendor’s ability to deliver high levels of peering for better quality, Peterson added. “How many peering points do they have in their network? Are the peering points local to avoid congestion?”
User Insight and Backup Plans
“Looking at jitter, packet loss, and latency on your network is entry-level analysis,” Peterson continued. “You need to look at the end-user experience. Are they having a clear and connected phone call with the lowest latency possible? Are they having good experiences on their apps, in their messaging? Are they able to share things globally, quickly? Are they able to establish a video conference quickly without having to figure out the one of seventy two video connectors that seem to exist in the world?”
An enterprise’s internal wireless network is the biggest weak spot RingCentral observes within its customer base, Peterson said, likely due to an apparent lack of talent in Wi-Fi administration and the know-how to deploy dense video and voice-ready Wi-Fi networks.
“We spend a lot of time helping companies with their Wi-Fi diagnosis and their Wi-Fi plans,” he said. “How do they plan to grow? How does hand-off work in their specific Wi-Fi hardware environment?” This is important because users can have a great experience at their desks, but if they walk our of the office and switch access points, it might be a terrible experience.
“And another thing-” Peterson said. “How many of you are familiar with an FOC date? It’s a Firm Order Commitment date. Carriers miss these dates all of the time. We recommend that our customers have multiple path strategies for connecting major branch locations. We have a lot of customers who use Internet as backup, even at locations with thousands of users, because Internet capability has really moved with the industry and is able to support these types of sites. Sometimes you’ll find that an Internet circuit can come in significantly faster than a WAN circuit. So look at your options and what the cloud starts to bring.”
Just five years ago, in terms of real-time applications on the network, you only had to worry about supporting voice and video, said Jamshid Rezaei of Mitel. “Now we also have all the applications that come with it, and all of the integrations such as tools for process flows and processing engines,” he said. “It’s very important that your provider can support all of these integrations, not just voice and video.”
Tata’s Elford summed it up this way: “It’s not just about the network, but about how you engineer it to have the most effective customer service or experience. You must enable a multi-cloud environment — whether it’s public, private, or hybrid — to provide application delivery that delivers the quality of service you need to get the benefits of cloud — consolidation and cost effectiveness — without losing control and visibility. You should not lose visibility of your network capability when you move to the cloud. You have to deliver on the promise of app availability, whether that’s on your mobile phone or sitting in a beautiful, immersive conference room. Either way, you expect to connect to that conference. Your provider should bring all this together seamlessly, giving you complete visibility.”
A deeper dive into factors around facilities, configuration, and failover that can complicate a move to SIP trunks.
An RFP that is issued at the proper time in the process, is supported by a thorough needs analysis, and clearly articulates the requirements of the organization, is an effective tool.
Carving out time to create an end user adoption strategy, and to execute it, will pay big dividends in the end.
Despite best planning efforts for UCaaS implementations, things aren't always what they seem when you get down to the nuts and bolts.