This week we have a case study. The company in question has just won a great contract to upgrade the software they provide to an industry sector across Australia and New Zealand.
Their supporting infrastructure is based on 16 servers - all four to 10 years old - for 60 staff and they need to upgrade the whole lot to run their project. They also have a really limited budget. What should they do?
What a fantastic opportunity to make use of modern technology to get set up for the next few years. First they must understand that supporting a server has a real cost of over $3500 per year and this is likely to cost more as it ages and causes interruptions to the productivity of their 60 staff members.
They should therefore allow a starting budget of $3500 per server that can be removed through consolidation. So if they can reduce from 16 to just three servers and shared storage solutions, the starting budget can be in the order of $45,500, just from the first year of savings plus reduced cost of management over the next few years, plus a whole lot more justified from productivity gains just to run their existing business.
Then add the real driver for this, which is the ability to develop and deliver their upgraded software product on a reliable & stable platform, getting the best possible productivity and lowest risk of downtime or data loss during the new project.
So if you are going to spend in excess of $100,000 what choices are there for you?
First of all we will be moving from single core processors to quad core processors, so the processing capability of each of the new servers is huge compared to what they had. As a result of this we can expect to run multiple servers in a virtual state on shared hardware. If we assume the whole lot will run on three dual processor servers with a virtual layer separating the operating system from the hardware we can be pretty confident that the eight processor cores in each box (24 cores in total) give us a lot of spare processing power.
Now we add shared storage via a storage area network or two. I suggest a mix of high speed drives for the line of business applications and transactional databases and some slower, less expensive drives for the offline systems and archived data.
Now with the right design and the right equipment we can build a system where each of the servers required can be dynamically shifted from box to box, test servers for the development can be run up on demand and then parked on the cheaper storage for use again later in the development cycle.
We can clone servers for testing without calling for new hardware and we can avoid downtime in the business if any of the physical servers fail. There is also the added advantage that this environment allows a test server to be run up in as little as 15 minutes, so no delays waiting for a spare physical server to be found.
This will also reduce the power consumption by over 50% and reduce the physical space requirements by over 70%, offering further cost reduction.
All this can be achieved with a 3% productivity gain across their existing 60 staff so, in fact, even though the business driver was the new project, the cost justification is there anyway. Why put it off until a big project comes in?
Their supporting infrastructure is based on 16 servers - all four to 10 years old - for 60 staff and they need to upgrade the whole lot to run their project. They also have a really limited budget. What should they do?
What a fantastic opportunity to make use of modern technology to get set up for the next few years. First they must understand that supporting a server has a real cost of over $3500 per year and this is likely to cost more as it ages and causes interruptions to the productivity of their 60 staff members.
They should therefore allow a starting budget of $3500 per server that can be removed through consolidation. So if they can reduce from 16 to just three servers and shared storage solutions, the starting budget can be in the order of $45,500, just from the first year of savings plus reduced cost of management over the next few years, plus a whole lot more justified from productivity gains just to run their existing business.
Then add the real driver for this, which is the ability to develop and deliver their upgraded software product on a reliable & stable platform, getting the best possible productivity and lowest risk of downtime or data loss during the new project.
So if you are going to spend in excess of $100,000 what choices are there for you?
First of all we will be moving from single core processors to quad core processors, so the processing capability of each of the new servers is huge compared to what they had. As a result of this we can expect to run multiple servers in a virtual state on shared hardware. If we assume the whole lot will run on three dual processor servers with a virtual layer separating the operating system from the hardware we can be pretty confident that the eight processor cores in each box (24 cores in total) give us a lot of spare processing power.
Now we add shared storage via a storage area network or two. I suggest a mix of high speed drives for the line of business applications and transactional databases and some slower, less expensive drives for the offline systems and archived data.
Now with the right design and the right equipment we can build a system where each of the servers required can be dynamically shifted from box to box, test servers for the development can be run up on demand and then parked on the cheaper storage for use again later in the development cycle.
We can clone servers for testing without calling for new hardware and we can avoid downtime in the business if any of the physical servers fail. There is also the added advantage that this environment allows a test server to be run up in as little as 15 minutes, so no delays waiting for a spare physical server to be found.
This will also reduce the power consumption by over 50% and reduce the physical space requirements by over 70%, offering further cost reduction.
All this can be achieved with a 3% productivity gain across their existing 60 staff so, in fact, even though the business driver was the new project, the cost justification is there anyway. Why put it off until a big project comes in?
If you need advice on your IT talk to Combo. Get it right the first time. We can help your business Cut Costs and Get Fantastic Service on IT. Don't spend money in the wrong places, invest in the right technologies for your business.
David Markus is the founder of Melbourne's IT services company Combo. His focus is on big picture thinking to create value in IT systems for the SME sector.
No comments:
Post a Comment