My very favorite quote on speed and measurement comes from
Alice in Wonderland:
"Now, here, you see, it takes all the running you can
do, to keep in the same place. If you want to get somewhere else, you must run
at least twice as fast as that!"
(Through the Looking Glass)
Interestingly, this passage actually relates very nicely to how vendors
measure latency in their networking devices. When
we read latency figures, we logically assume that the number represents the length
of time it takes for a packet to enter the switch and come out the other
side. True enough, as far as it
goes. But there are lots of ways to
measure the speed of a packet as it traverses a switch. And there are lots of questions that should
be asked in order to really understand what was measured and how. An interesting new white paper that details
these questions and provides additional insight into what “latency
measurements” really mean can be found here.
So why settle for microseconds when you can have
The more work you can get done in each second, the more
productive you are—especially if you are in the financial services or high-performance
computing industries. Switches that deliver
transactions in terms of nanoseconds are 10 times more productive than their
If speed in nanoseconds is the goal, how do you find the
fastest switch in this category? There
are several dimensions that should be taken into consideration: latency, jitter, and multicast.
Latency considerations include packet size and
throughput. How fast can a switch
transmit each packet size at wire speed and not drop anything? To really understand how a switch behaves
under stress, you need to test all ports at the same time using 100% unicast
traffic throughput in a fully meshed pattern using RFC 2544 and RFC 1242 (for
cut through switches).
It is very easy to go fast on the freeway when there are
no other cars. So you need to look at
how fast a switch transmits at wire speed through all ports with full traffic
load. Anything less than less than 100%
line rate traffic and full mesh to all ports is wimpy.
I’ve pointed out that production data centers have
five different variations, and I’ve covered four of them in previous blogs.
Today, I’ll talk about the fifth—hosting data centers.
Hosting production data centers come from the
classical hosting services market. These
providers started by hosting servers for customers who did not want to maintain
their own server farms. Unlike having
your content in a cloud, you know exactly where your content is—it is in the
servers that the hosting provider manages for you.
I’ve mentioned before that there are different types
of production data centers. Cloud service providers are production networks
which contain public and private content.
These are the newest type of production data center. They are different in that you have to pay to
put things into them (as opposed to paying to get things out). Cloud providers give you the same type of
dynamic data center scaling capability as the largest content companies in the
world. Your content is virtualized and
moved around in a “cloud.” Unlike other types of Content Data Centers, these
will provide Service Level Agreements (SLAs) for your content. They have
additional requirements for traffic separation and security that the public
content providers do not offer. That is
because they need to keep your content separate from everyone else’s.
Content can be roughly defined as information. It exists as words, images, video, music,
voice, TV and any form of media. Over-the-Top Content data centers are
concerned with throughput and scale.
These data centers house public and private content which grows
continuously. Video, photos, news, text
on every subject known to humanity produced in every language used by mankind (e.g.
Facebook). These data centers account
for much of the growth in servers.
Content data centers are challenged by space, cooling and power. These data centers can take the full output
from nuclear power plants and still need more.
Over the past several weeks, I’ve been talking about
how data centers can vary significantly and discussing the technology decisions
companies need to make in order to help them scale in the future and be
successful. Today, let’s look at Production data centers. These are profit oriented. They require leading edge solutions. They purchase new technology to get the
latest and greatest in performance. The most demanding of these is the financial
and High-Performance compute (HPC) data center.