IT Managers’ Journal tiene un magnífico artículo escrito por el responsable de IT de una empresa de más de 300 usuarios con 6 diferentes sucursales en USA, y de cómo ha sido su experiencia en la migración de Windows a Linux a nivel de servidores, y los ahorros que el mismo encontró en la misma. En la conclusión de su artículo, Ryan Benner lo tiene claro. Linux permite ahorros de más del 50% del TCO
Aconsejo a todos los admnistradores, CFOs, CEOs y personal IT que no han estudiado a Linux, a comenzar a mirar este software como una solución viable para los precios de otros sistemas. Puedes ahorrar la mitad de los costes de administración, de licencias y de hardware ; puedes incrementar la seguridad y el uptime; y no tedrás que hacer migraciones masivas cada pocos años para mantener las opciones de soporte
Migration from Windows to Linux saves thousands
Wednesday January 14, 2004 – [ 08:00 AM GMT ]
Topics: Management & Administration , Network Management & Administration , Software , Personal Electronics , Networking
By: Ryan Benner
Nearly three years ago I rebuilt my company’s corporate network, comprising six geographically dispersed offices and approximately 300 users, using a budget smaller than what most system administrators and IT managers make in a year. Our migration to Linux servers and software was a success, and offers a lesson for other administrators.
When I began, I had quite a mess on my hands, and I don’t mean the kind that our company, a maker of water treatment products and services, was accustomed to cleaning up. I encountered problems such as not enough memory for the servers to run their services and not enough hard drive space to accommodate growing data needs. We did not have enough licenses to support our company user base, and Microsoft was contacting us to request we update our licenses and become compliant. I was also experiencing
unplanned down time at least once a week, and suffering with an inadequate backup system, a shaky T-1 line, and no security — not even a firewall. Our company lacked a single corporate email system; out of 300 users, approximately 50 were using Exchange Server; the other 250 were using AOL or another ISP.
After spending half the budget simply getting the network to a functional level, I started to visit the upgrade feasibility. I needed a very simple solution. I started with trying to maintain a Windows server network as they had originally.
I submitted a proposal for $175,000 to upgrade three servers, get the company licensing compliant, and upgrade from the older Exchange to the latest version running on Windows 2000. The estimated price also included some consulting help, but the consulting help did not have a not-to-exceed clause.
I was going to need 300 Microsoft CALs (client access licenses) at around $35 each, times two servers. You might wonder why only times two and not times three, since we needed three servers, but not all 300 users would be accessing all three servers. The total came to about $21,000 for the Microsoft CALs for the two servers everyone would need to access. I needed 40 more CALs for corporate users on internal servers, times three for the three internal servers we needed. I couldn’t completely remove all of the old servers, since the file, print, and data needs were too great to try and slim down to only three servers. That made about $4,200 for the internal users located in our corporate headquarters. The total came to $25,000 just for the users to access the servers. That’s a lot of licensing just to access the servers, and I hadn’t hit on the new hardware yet.
After I’d finished resolving the CAL problems and making the company compliant, I needed to go back and look at the hardware. The company had nine white box servers which could barely handle their current data needs, let alone support any kind of Active Directory rollout. With the help of a consulting company, we narrowed down our hardware choice to HP NetServers. In order to have enough drive space, memory, and processor power to facilitate what was required, each computer would cost about $10,000, and the system that would run Exchange topped $20,000.
Now we were up to $40,000 for the hardware plus $25,000 for the server licensing.
Next, we needed to get licensing for both Microsoft Exchange and Outlook. Most employees already had Outlook installed on their systems, so I purchased 20 more Office licenses for about $495 each, an approximate total of $10,000. Exchange user licenses ran about $65 per user; with approximately 300 users, they would cost approximately $20,000. That brought the combined total to $85,000 for licensing fees and hardware costs.
Now the implementation costs. The consulting group we were going to work with quoted us almost 400 hours of work at a rate of $225 an hour, for a total of $90,000 for most of the training and the configuration.
Needless to say our CFO was not happy about the amount. I agreed with him. So I walked down to our VP’s office and asked him, «Since the cost for Microsoft is so high, how about taking the plunge and looking into Linux?»
His first question was, «What is the risk of switching from Microsoft to Linux?» followed quickly by «Will it be cheaper?» and «Who will support the system?» I told him the price would be substantially lower, so our out-the-door costs in a year would be insignificant and we could migrate back if we needed to. To answer the support questions, I cited the huge Linux community.
I got the approval I needed to do a comparative estimate for a Linux product, but my search for a Linux solution did not come without conditions. In the event of my untimely demise or if I outlived my usefulness with the company, I needed to teach someone the system.
I contacted Cobalt, a Linux hardware vendor that would soon be bought by Sun, which had provided our company with a Web server, and started to look at its products. Cobalt’s Raq 4 hardware used a simple browser-based administrative interface for setting up users and groups, and it used Samba for file and print services. The cost was about $1,500 per server. If I decided to go the Cobalt route, I would need to purchase five servers for a total of about $7,500. That sounded better than approximately $40,000 for hardware for the Windows machines.
To bring the company into a global email situation I first looked at Emumail, but discovered it would need a lot of programming to get it to work for us. With the help of friend and consultant Jon Traxler of Netraxx, we did some searching for a Web-based POP3 email system. We came up with using VirtualTek’s Joydesk, which cost about $2,500 for 300 users. That brought combined hardware and software cost to about $10,000.
The migration of data was a simple drag and drop. The only thing left was consulting, so again I leaned on Netraxx’s Traxler. His fees were about $100 an hour at about 200 hours, for a total of $20,000. That made our total cost around $30,000 to change from Microsoft to Linux on the server side, a savings of approximately $145,000 over the cost of upgrading.
Since I’d saved so much on the cost of migrating from one operating system to another, I decided to purchase a new backup system from Penguin Computing, running Red Hat 7.1, for a cost of about $1,700. I then upgraded my Computer Associates ARCserve backup to the Linux 7.1 version for approximately $1,500, and I purchased a PowerStor L200 DLT drive for approximately $4,000.
I also purchased a new test server from Qli for approximately $1,900. Though it came with SuSE Linux, I purchased Red Hat Professional for $149 and I installed it on the server. That gave us seven Linux servers for a total cost of $13,000.
After having the system up and running for approximately 3 years, I had to purchase a new email server. Joydesk proved buggy and crash-prone at our site. I am currently getting ready to migrate email and collaboration over to Oracle Collaboration Suite, which will cost us approximately $60,000 for 300 users. All in all, the system is showing about 99.999% up time on everything except for the email software. I have very few problems, which I equate with the infomercial Set It and Forget It. Switching over to the Linux servers that act as file and print servers network authentication servers has opened up my work time for research and other projects that contribute to company growth.
I did keep one NT server for our lab database and HR and payroll software. Do I think a company can become fully free of Microsoft, its high prices, and poor security all the way down to the desktop and applications? I do. My next projects include desktop systems, and sometime soon I hope to migrate from an AS/400 server to a better accounting program that runs on Linux.
I advise all the administrators, CFOs, CEOs, and IT personnel who have not taken a look at Linux to start to look at the software as a viable solution to price-gouging of other systems. You can cut administrative costs, licensing costs, and hardware costs by half; you can increase security and up time; and you won’t have to make major migrations every few years in order to maintain your support options.