Hardware Reference
In-Depth Information
The conventional WSC wisdom is to run other valuable tasks during periods of low activity
so as to recoup the investment in power distribution and cooling. A prime example is the batch
MapReduce jobs that create indices for search. Another example of geting value from low util-
ization is spot pricing on AWS, which the caption in
Figure 6.15
on page 458 describes. AWS
users who are flexible about when their tasks are run can save a factor of 2.7 to 3 for compu-
tation by leting AWS schedule the tasks more lexibly using Spot Instances, such as when the
WSC would otherwise have low utilization.
Fallacy Replacing All Disks With Flash Memory Will Improve Cost-performance Of
A WSC
Flash memory is much faster than disk for some WSC workloads, such as those doing many
random reads and writes. For example, Facebook deployed Flash memory packaged as solid-
state disks (
SSDs
) as a write-back cache called Flashcache as part of its file system in its WSC,
so that hot files stay in Flash and cold files stay on disk. However, since all performance im-
provements in a WSC must be judged on cost-performance, before replacing all the disks with
SSD the question is really I/Os per second per dollar and storage capacity per dollar. As we
$2.00/GByte versus $0.09/Gbyte.
Narayanan et al. [2009]
looked at migrating workloads from disk to SSD by simulating
workload traces from small and large datacenters. Their conclusion was that SSDs were not
cost effective for any of their workloads due to the low storage capacity per dollar. To reach
the break-even point, Flash memory storage devices need to improve capacity per dollar by a
factor of 3 to 3000, depending on the workload.
Even when you factor power into the equation, it's hard to justify replacing disk with Flash
for data that are infrequently accessed. A one-terabyte disk uses about 10 wats of power, so,
using the $2 per wat-year rule of thumb from
Section 6.4
,
the most you could save from re-
duced energy is $20 a year per disk. However, the CAPEX cost in 2011 for a terabyte of storage
is $2000 for Flash and only $90 for disk.
6.9 Concluding Remarks
Inheriting the title of building the world's biggest computers, computer architects of WSCs
are designing the large part of the future IT that completes the mobile client. Many of us use
WSCs many times a day, and the number of times per day and the number of people using
WSCs will surely increase in the next decade. Already more than half of the nearly seven billi-
on people on the planet have cell phones. As these devices become Internet ready, many more
people from around the world will be able to benefit from WSCs.
Moreover, the economies of scale uncovered by WSC have realized the long dreamed of
goal of computing as a utility. Cloud computing means anyone anywhere with good ideas
and business models can tap thousands of servers to deliver their vision almost instantly. Of
course, there are important obstacles that could limit the growth of cloud computing around
standards, privacy, and the rate of growth of Internet bandwidth, but we foresee them being
addressed so that cloud computing can lourish.
Given the increasing number of cores per chip (see
Chapter 5
), clusters will increase to in-
clude thousands of cores. We believe the technologies developed to run WSC will prove useful
and trickle down to clusters, so that clusters will run the same virtual machines and systems
Search WWH ::
Custom Search