-Don't leave the lights on! One reason we take energy consumption seriously is because we are directly billed for it. If one leaves the heat on high over night the effect is noticeable on the next bill. Yet as granular as cloud computing billing can be in terms of resources and quality of service, we have limited motivation to investigate energy consumption in the cloud because cloud customers cannot necessarily realize savings. Proxies for quality of service such as lower performance CPU allocations can be used but at no point does a user see a bill listing energy consumption. Furthermore the difficulty in billing energy consumption of virtualized services is non-trivial and indirect. When many VMs share the same host, attribution of energy consumption becomes difficult. When many hosts are in the same datacenter attribution of cooling costs become difficult as well. Thus due to the direct and indirect costs of running a cloud, and the sharing of resourcing pricing cloud energy consumption is difficult and typically not done. We argue that until energy consumption of hosted computers, VMs, and cloud services is pushed down from the cloud provider to the cloud consumer, datacenters will continue to consume massive amounts of energy to provide software services. When cloud end-users have to pay for energy consumption they will consider optimizing energy consumption. Once energy consumption in the cloud is a bill line item, energy consumption will become a first class performance non-functional requirement of software.
We propose a lightning talk to be delivered within 5 to 10 minutes at RE4SUSY 2018. The lightning talk will cover the topic of lack of demand for software sustainability features until cloud services start billing for energy consumption.
During a discussion with some unnamed industry partners we asked about energy consumption, they responded that they were concerned, but until they could put a price on energy consumption of their software systems, they were going to ignore it. Why bother to optimize or even plan for energy consumption when you had immediate pressing concerns from requirements and non-functional requirement alike including features, stability, reliability, and cost of cloud services. Typically cost of cloud services already includes energy consumption and it is not a line item—thus optimizing for energy consumption does not have a clearly observable effect on the cost. Without insight into, feedback about, or the cost pressure of energy consumption, it will not be optimized for or even requested for from customers.
Energy consumption of software systems and services is a
grave concern as data-centers now account for a significant
portion of the worldwide energy consumption (2-3% [
Cloud providers face tough decisions regarding energy consumption as elasticity and availability are buffered by the high cost of idle machines in the datacenter. High utilization is a goal, achieved by slicing servers up into VMs and containers. Yet high utilization has many other costs, heat, maintenance, energy, poorer per VM performance, are just a few. Heat requires further cooling thus consuming even more energy.
While it is easy to slice up the resources of a server and
add it to a cloud, it is hard to attribute energy consumption
to each particular VM, container, or service on a cloud
server, although there is much research in the area [
We argue that nothing will motivate cloud end-users to optimize for such an invisible externality until it becomes a reality on their cloud services bill! Until the motivation for sustainability can be expressed monetarily on a bill for services, there will be few requirements communicated from customers to address software energy consumption of their cloud services. Given the current trend in cloud billing where giants such as Amazon rate CPU time and even engage in CPU credit systems with fine granularity there is likely pressure to bundle the cost of the externalities of a data-center into the cloud costs as well. The closer the customer can get to paying for exactly what they using the easier both parties can plan and expand. Thus we argue there is pressure on the part of cloud providers such as Amazon, Microsoft, and Google, to ensure that they can push all of their costs to the enduser proportionally. The value in not direct billing is that the customer assumes it is accounted for, but what the customer cannot see or measure, they will not optimize for. A customer who uses non-optimal instance VMs has no incentive to avoid this behaviour—for example low memory VMs coupled high numbers of CPU cores waste memory as they deny the unused memory to other VM instances who cannot be scheduled to that machine until the cores are freed up.
We argue that the incentives for sustainability should be, and eventually will be, pushed down to the cloud-service end-user. Until that point, few will prematurely optimize for energy, and sustainability when they can directly optimize for other costs.
Energy consumption can come at a high cost, Song et
al. [
As data-centers are prone to deploy near where people
actually live—in order to reduce latency—they often prone
to rely on energy produced in the local area—for instance
in Edmonton, Alberta much energy is provided by coal fired
power plants which have signficant environmental and health
impacts. Often this energy has significant carbon emissions
and thus data-centers energy will also be taxed accordingly
with carbon-taxes. We argue that much like carbon-taxes,
enduser billing of software energy consumption will promote
reduced energy consumption or at least sustainable energy
consumption much like economic models of carbon tax predict
will happen with emissions and emission producers [
If end-user billing of software energy consumption of cloud
services became possible then those hosting services would
have another dimension of performance to optimize for. Many
other researchers have argued that users might be willing to
trade off one concern to improve a performance concern such
as network bandwidth or energy consumption [
Thus we have argued for the need for software energy
consumption billing in cloud computing services—we believe
it is inevitable as price and quality of service is a constant
concern and competitive for cloud services. We believe that the
advent of itemized billing for energy and other sustainability
concerns will prompt cloud-services end-users to care and
optimize for energy consumption concerns. Conversely, as
Knuth stated, “pre-mature optimization is the root of all
evil” [
Cloud providers will have to convince users to adopt a
billing scheme based on energy consumption estimation as
it cannot be measured directly. Direct measurement is not a
large barrier as the governments of world propose taxation
and incentivization strategies to address sustainability issues
and many unmeasurable externalities. It is unlikely that
private service providers cannot enact what governments have
successfully deployed [
Cloud providers should bill customers directly in order to motivate requirements of sustainability and foster further innovation in software requirements and software design relevant to software energy consumption and sustainability.