Executive Summary ================= Get the company to profitability sooner by saving costs with out incurring sacrifices. Cost savings would be achieved by reducing the company's dependence on foreign oil. By foreign oil, I of course mean proprietary software and support licenses as well as electrical power consumption. IT Cost Saving Plan Summary =========================== Via strategically converting various network services over to open platform, open source alternatives, without sacrificing services offered, a large cost savings can be achieved while also bringing the company to a more open platform compliant state. In other words, walk the walk, not just talk the talk. Plan Details ============ The company expends a great deal of money on software licenses for network services such as email, firewall, database, source control management, and issue tracking. All of these services enjoy particularly competent open source offerings, many of which are considered more competent and reliable than the proprietary offerings we pay high license fees for. In addition to these services, we are forced to pay for expensive software that allows us to access these non-open platform services from our non-windows platforms, of which we rightfully have a large number. I am referring to Citrix of course. If that wasn't enough, our email/calendaring service is so inefficient that we have to spend large amounts of money on special high performance machinery just to get it to run at a barely useable speed. Current Software list Yearly cost --------------------- ----------- Exchange $999.99 Citrix $999.99 Perforce $999.99 Clearquest $999.99 MSQL $999.99 Firewall software (?) $999.99 Replacement software list Yearly cost ------------------------- ----------- Exim4 $0.00* webcalendar $0.00* SVN $0.00* Bugzilla $0.00* Linux or OpenBSD based firewall/VPN $0.00* Under this plan, the software package Citrix is to be completely eliminated. The rest are replaced by open source software. Caveats ======= Some users might wish to continue to use their non-Windows platforms, but continue to use Windows applications via Citrix [grounds for termination?]. So that software may need to remain, but even so, a substantial cost savings can be achieved by dramatically reducing the number of seat licenses that need be purchased. A decent plan to switch email and calendaring over to their open source alternatives has to be laid out in order to avoid disruption. [Overheard in company hallway: how did we get anything done before M$Exchange/Calendar? Answer- the same way we do now.] A beta trial with a small number of employees is advisable. *Truth in advertising ==================== The open source paradigm is a much different one than proprietary software usage model. In order to be successful, the right model has to be followed. The primary difference is support. Open source software is largely self supported. This generally means at least one additional employee in IT. Even if self supporting does not amount to any dollar savings over purchased support contracts, our economy and our corporate karma are greatly helped by employing a person rather than paying these fat companies their over priced support fees. And there is another advantage: we get the software tailored to our desires and whims, customized at will. Something that can never be achieved with proprietary software. Email/Calendaring switchover plan ================================= Current (non-upgraded hardware) is probably more than sufficient to run email and web calendaring services at a very high performance rate. Therefore, we can keep the current machinery and execute the following plan: 1. Export email database in a form that is importable by exim4. This will take a bit of research but is quite doable. 2. Import email to beta trial server. 3. Set up webcalendar and integrate it to email (ical). 4. Take a small number of sophisticated volunteer users to use the new services for a month, providing feedback to IT. Power Saving Plan Summary ========================= Phase out low performing, high power consumption hardware, and replace with higher performing, very low power consuming hardware. Plan Details ============ The walls of our company are plastered with articles about greening of data centers and machine rooms, but in side our machine rooms, we ourselves are completely ignoring the issue. It's time to realize a huge potential cost savings by reversing that situation. The largest piece of low hanging fruit in this area is the number of low performing, high power consuming single and dual Xeon systems, both in the lab and out. These systems typically consume several hundred watts while sitting idle, yet their performance is somewhat marginal for all that heat generated. Gradually replacing these systems with lower power consuming systems made with low power offerings by AMD or Intel would net us a very large savings, both in direct electrical costs as well as cooling and power distribution and UPS costs. A typical single 2.8HGz Xeon consumes 169 watts, and that doesn't include it's chipset, which requires an extra chip that Opteron/Athlon64 CPUs don't require. Low power Athlon x2 processors with greater performance and lower power consuming DDR2 memory consume as little as 45 watts. More than double the computing performance, at 25% of the power consumption. Multiply that by the number of such client system in our lab alone, and we have achieved a very large power savings. Such systems also are very cost effectively priced as well. High power consumption lab systems would be gradually replaced with new power saving units. A budget of 2-3 systems per month should be set in order to not overload lab personell or disrupt client usage by QA and development staff. Development and QA workstations can also be replaced at a 1-2 unit per month rate, possibly higher depending on IT availability. The cost savings and performance increase will be sizeable in only three month time frame. Higher performance directly correlates to higher efficiency of engineering and IT staff members. Savings Estimate ================ Such an estimate involves a high degree of speculation, but half a million dollars a year if both plans are implemented is not out of the question. Possibly even higher.