Written by Jay Botelho, Director of Engineering at LiveAction
According to a recent
survey of IT professionals, 42% report spending too much time on
troubleshooting tasks. Additionally, they cited the cloud as the second most-common
place in which they encounter significant IT issues. These findings aren't too
surprising for most in networking - a lack of visibility into cloud
infrastructure and applications forces network operations (NetOps) teams to fly
blind, unable to predict issues proactively or troubleshoot quickly. Ultimately,
this negatively impacts user experiences and stokes frustration among employees
and customers. To access the level of visibility
necessary to prevent or mitigate these issues, IT and NetOps need to prioritize
monitoring and incorporate the right network management solutions into cloud
deployments from the beginning. This allows NetOps to keep business-critical
applications running smoothly as they transition to the cloud, pinpoint the
root cause of issues with cloud applications, negotiate effectively with
service providers and reduce finger-pointing while troubleshooting.
The exact details of this process depend on the type of
cloud deployment you're dealing with. Let's walk through two examples step-by-step
to illustrate these best practices.
Software as a Service
Organizations moving from applications running on a local
server to a Software as a Service (SaaS) provider (i.e., moving from a local
Microsoft email server to Office 365) have limited options for visibility. IT
teams can't install monitoring software in Microsoft or Salesforce, so these
applications can often seem like black holes or networking blind spots. In the
past, many businesses gave up trying to monitor SaaS applications, but now that
IT departments are increasingly held accountable for their performance, they
must find workarounds. Luckily, there are ways to get at least partial
visibility into this traffic. Here's how:
- Before transitioning to a SaaS application, use
a network analysis tool to categorize the performance of the application running
in the data center. Measure how much traffic is flowing back and forth, network
response times, application response times, etc.
- Write these specifications into your contract
with the SaaS provider. Make it clear that you expect your users to get the
same application response times as they did locally, or better yet, see an
improvement.
- After moving to the SaaS application, place a
monitoring point at the firewall or connection to the internet, and monitor the
round trip to that SaaS application. Measure the overall delay and compare to
before the cloud migration.
- Monitor this over time. You may not be able to
troubleshoot the SaaS application directly, but with this method you can see if
performance is changing over time and will be aware of any significant degradation.
- Using the data collected, negotiate with your
SaaS provider as necessary. For example, if you have data from the past six
months showing the average application response time for Office 365 going from
200 ms to 500 ms, and the SLA says 200 ms, you can ask Microsoft for a refund and
to address the performance issues so you're getting what you pay for.
Infrastructure as a Service
Conversely, organizations using infrastructure as a service
(IaaS) that have set up their own cloud applications and architecture have more
options when it comes to gaining detailed network visibility. Here are the key steps
any organization going this route should follow to ensure maximum visibility:
- Follow steps one and two in the SaaS example
above - dig into the current application performance on the local server and
set specifications for performance with your IaaS provider.
- Place a monitoring point in your cloud
infrastructure to monitor your north-south traffic. This allows you to track
network and application response times from your users to your cloud
architecture through the Internet Service Provider (ISP). This step is very
important because it allows you to determine if an issue with an application is
the result of the application itself, a mistake in how you've set up your cloud
architecture, or an issue with the ISP.
- This may be all that some companies need to
monitor. For others, especially those with more complex cloud deployments (such
as moving an entire data center to the cloud), it makes more sense to expand
and add additional monitoring points. This enables them to monitor the
east-west traffic between different applications and between applications and
databases within your cloud deployments, allowing for more detailed and
effective troubleshooting. If a NetOps team determines that an application
slowdown is the result of an issue within the cloud, they need this level of visibility
to pinpoint the error.
Remember, when using IaaS, IT departments are simply renting
"gear." It's up to them to use it correctly and troubleshoot their own design.
While you can rightly take your SaaS provider to task for delivering poor
service, you can't shout at AWS if their cloud-hosted applications aren't
running as fast as you need them to.
What happens in the cloud doesn't stay in the cloud.
Regardless of the visibility challenges, cloud issues can have a major impact
on business operations. Organizations should ensure that they have a unified
Network Performance Monitoring and Diagnostics platform
that offers full end-to-end visibility into all fabrics of the network,
including the cloud, to speed troubleshooting and put the data they gather from
the monitoring methods explained above in the context of the entire network.
Overall, the best cloud monitoring setup depends
on each individual organization's unique needs, but all businesses should at
least monitor traffic in and out of the cloud to their corporate users. Whether
using SaaS or IaaS, every IT and NetOps team needs to pinpoint the location of
application issues quickly and efficiently and be able to verify that they are receiving
the performance they're paying for. With the techniques explained above, IT
will be well-equipped to support business-critical applications as they move to
the cloud and keep the business running smoothly through ongoing network
transformation initiatives as well.
##
About the Author
Jay Botelho is the Director of Engineering at LiveAction