Green From the Bottom Up
There has been a lot of media attention over the last 24 months regarding companies and the “Greening” of their business models.
Typically, from an engineering standpoint, eco-responsibility in technical products affects two areas: Manufacturing, in the form of ROHS (Reduction of Hazardous Substances) standards, and in the design - for reduced power consumption.
After Sun Microsystems launched its first generation Niagara servers (T-1000 & T-2000) - which were fully ROHS certified - we saw Pacific Gas & Electric offering rebates to utility customers who were upgrading older servers to the new lower power consumption models. This is a prime example of eco-responsibility crossing industries. Talk about team work! 
Dell has spent thousands of hours working on power efficiency in its new line of M Series Blade Servers. They’ve also extended their power efficiency efforts into the design of their consumer-based products, with twenty-one of their Optiplex, Precision, and Latitude products certified for Energy Start 5.0 compliance. They already have 2 servers that are Energy Star-compliant, and are working on more. 
Hitachi Data Systems is revolutionizing eco-responsibility in the Enterprise Storage arena. All of their current storage products are fully ROHS-certified, placing less strain on the environment after the usable lifetime of a product. They’ve also done a lot of work with managing power consumption. If you’ve ever managed a SAN - Enterprise class or not - you know that keeping several, to several hundred hard disks spinning at all times doesn’t come cheap on the power front. By combining intelligent disk controllers with innovative power management software, they’re able to significantly reduce the power requirements of their storage systems.
Without question, a global leader in engineering for power efficiency and eco-responsibility is Google. Google recently unveiled their standard server used in all their facilities. By using such techniques as on boards batteries, rather than large UPSes, they’re able to achieve 99.9% efficiency as compared to most large UPSes which range from 92-95% efficiency. With a scale like that, a lot of money and power will be saved. Read more about the level of scale in which Google operates 
This is not by any means a comprehensive list of the Green IT initiatives out there. Due to legislation in the EU, California, and other parts of the world, there are really no options for computer manufacturers who like to sell their products to avoid the ROHS side of manufacturing. And, while it’s not legislation, it’s common knowledge that a large percentage of existing datacenters are simply not able to provide additional power. This leaves manufacturers no other option but to design servers and storage systems with ever-greater power efficiencies.
All of these innovations are great. Good for the CFO … good for the CIO … and good for the PR department. But how do these benefits really translate for the average Sysadmin?
There are several factors at work in the modernization of the datacenter environment. We have Green Shift, which is the rise of all this Eco-Responsibility in computing - the lower power consumption servers, blades, witches, routers, and storage systems. But there’s also a Red Shift going on, which is the exponentially increasing storage and compute requirements that companies have. The pure amount of information generated each day has to be stored somewhere, and has to be processed somehow.
These factors, combined with new technical features in servers like hypervisors and virtualization , have led to some very interesting consolidations of expansion. That sounds like a contradictory statement, however, if you are an IT Manager responsible for a moderate environment of say 40 servers, 4-6 storage arrays, and the corresponding requirements for switches, routers, firewalls, and backup infrastructure, how do you handle the increasing demands for the services you provide, while at the same time respond to the increased pressure from green legislation and cost constraints in a down economy, not to mention power constraints in the datacenter, to achieve your goals, save your company money, and prove your commitment to Green IT to your customers and/or investors?
Well, you could start by consolidating your 4-6 legacy dumb storage arrays onto a smarter, larger, more power-efficient SAN. That would cut your existing power requirements from 8-16 circuits down to possibly 4. That’s a pretty big power savings right there, given the cost of electricity in datacenters, also taking into account the cost of keeping your gear cooler.
Maybe you also take your 40 older, single or dual CPU servers, with 2-8GB of RAM, and 2 internal disks, and consolidate them using any number of virtualization technologies onto 4-8 of these newer servers.
To be Sun-centric for a moment, let’s say your 40 existing servers are Sun V210 servers with 2 x 1.34 GHz UltraSPARC-III processors, and 16GB of RAM, and 2 internal 146GB hard disks, mirrored, running Solaris 9. That would give you a total of 80 1.34ghz CPUs, 640GB of RAM, and 5.8TB of onboard storage. 
You could consolidate all of those servers onto 4 Sun T5240 servers with resources to spare. Without getting too far into the capital expenditure required for the purchase of the T-5240 - which loaded up is not a cheap purchase - you’re going from 40 servers to 4. From 80 power circuits to 8. From 80 compute threads to 512. From a maximum RAM capacity of 640GB to a maximum RAM capacity of 1024GB. From a total internal storage capacity of 5.8TB to a total internal storage capacity of over 19TB. 
Even with all the added RAM and Disk resources, you’re still saving a ton of money on power. Your CFO will be even happier at the reduction in cost of your relevant support contracts, too!
From the Sysadmin perspective, some internal tools may need to be updated to accommodate multiple managed servers running in parallel on the same hardware, but in general, there is less hardware to maintain, less devices to fail (although a failure can be more critical in nature if a single physical server failing can take down a large percentage of your environment). With some options like Xen, and VMWare and some of the other virtualization Technologies, there are ways to do both live migration and cold failover migrations which can eliminate - or at least minimize downtime - in the event of a failure or a required maintenance to the host system.
If you’re using Solaris 8/9 Containers on Solaris 10, you have another added benefit as a Sysadmin in that you can now use DTrace to peer from the Global Zone into processes running in the Solaris 8 or Solaris 9 containers. That can have a big impact on the quality of performance tuning systems, or the ability to quickly troubleshoot problems. 
From the bottom up, the Greening of IT really provides a wealth of opportunity for the front line Sysadmins, easing our jobs, and reducing our workload, while providing maximum growth potential and a significant reduction in costs for our management to be happy about from the top down.
As a company providing services ranging from pure colocation, to fully managed cloud computing environments - and everything in between – Carpathia Hosting utilizes a wide variety of these technologies ourselves, and on behalfof our customers. We provide an entry-level virtual private server offering based on the Parallels platform. We also have VMWare enterprise-level virtualization as another option, and a cloud computing platform that leverages Xen in combination with some third party software to enable some significant management improvements over base Xen. We have customers using Solaris Containers, and customers using Solaris Logical Domains. We continue to consolidate our own internal environments, and we are fully committed to supporting our customers in implementing their own eco-responsible environments and achieving their goals as relate to the reduction of their impact on the environment in the course of their business.
Jumping on the Green IT bandwagon is pretty much a requirement these days due to the aforementioned legislative activities, and the fact that almost all manufacturers of technical equipment have very stringent ROHS requirements to adhere to. Sometimes, the only way to get power in a datacenter is to remove equipment - but it’s also the right thing to do - and in the end, it really does make our job easier, and provides a real monetary value to our customers.
If you are a Sysadmin, what impact, if any, has Green IT had on you or your job?