Over a million developers have joined DZone.

Amazon EC2 Under Fire About Network Issues

DZone's Guide to

Amazon EC2 Under Fire About Network Issues

· Cloud Zone ·
Free Resource

Learn how to migrate and modernize stateless applications and run them in a Kubernetes cluster.

Amazon Web Services have led the charge for IaaS cloud computing over the last few years with RDS (Relational Database Service), Small Instances, and instance bidding.  Their mantra has been: 'Give us everything you've got.  We can handle it!'  However, several customers have argued that recent network issues prove Amazon's infrastructure has limits.

Reports of higher ping latencies started late in December and have gotten much worse since the start of 2010.  According to some users, many instances don't seem to be performing nearly as well as they should.  Many customers have traced the problem back to network latency and over-capacity issues, but Amazon says this is not the issue:

“We do not have over-capacity issues,” said Amazon spokesperson Kay Kinton. “When customers report a problem they are having, we take it very seriously. Sometimes this means working with customers to tweak their configurations or it could mean making modifications in our services to assure maximum performance.”

Ping trace example from a AWS user experiencing high latency

64 bytes from icmp_seq=22 ttl=61 time=0.868 ms
64 bytes from icmp_seq=24 ttl=61 time=1350 ms
64 bytes from icmp_seq=25 ttl=61 time=4183 ms
64 bytes from icmp_seq=27 ttl=61 time=2203 ms
64 bytes from icmp_seq=31 ttl=61 time=0.554 ms
64 bytes from icmp_seq=32 ttl=61 time=678 ms
64 bytes from icmp_seq=34 ttl=61 time=0.543 ms
64 bytes from icmp_seq=35 ttl=61 time=25.6 ms
64 bytes from icmp_seq=36 ttl=61 time=1955 ms
64 bytes from icmp_seq=41 ttl=61 time=809 ms
64 bytes from icmp_seq=43 ttl=61 time=2564 ms
64 bytes from icmp_seq=44 ttl=61 time=7241 ms

Over capacity or not, the latency issues have forced some customers into buying higher priced "High-CPU Medium Instances."  Amazon's new Small instances have become very popular since their creation.  However, their increasing latency for some customers may be a sign that Amazon can't handle the large volume of Small instances now running in their cloud.  The High-CPU Medium Instances reside in a different part of Amazon's infrastructure with higher resources, but even they have begun to have latency problems similar to the Small instances.  

Some companies have not experienced latency if they are using larger instances.  Customers using applications that mainly move or process data aren't seeing as many issues on Amazon's cloud as the applications that require low latency.  It's possible that the ICMP pings used to test latency are inaccurate because they receive lower priority.  However, many companies have ruled out a hardware problem on their end with extensive testing.  Cloudkick says the latencies of up to 1000ms that people are experiencing are, "definitely abnormal, and should never be encountered on healthy private network." 

Join us in exploring application and infrastructure changes required for running scalable, observable, and portable apps on Kubernetes.


Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}