We made a Freakin' Rocketship

We pared down the clients servers from 9 to 1.  That's right - they had 9 servers - yet only 10 staff.  They had been sold on having multiple redundant servers - and then backups for those servers.  So the first thing we did was turn off 6 immediately if only to save electricity.   

The ratio of servers to staff was no longer 1:1 but a more manageable 1:10, and we were hoping to combine the rest onto just 1.  A firm of 10 people should only have 1 server.  The normal staff to server ratio should be around 20/30 to 1, perhaps a little lower for Architecture firms.



we have no problems

We got the green light - and gave them a rough estimate based on a similar spec for another Architecture firm - so we should be fine.   But the prices dropped all over the place - so we went back with 3 options costing between $7k - $9k.

They decided to go with the more expensive option.  This took a little bit of convincing and some guarantees on our part.   As we're testing a new hybrid configuration, I wasn't sure if just how fast it would perform, so I promised that if they did not see a large performance increase - I would refund the difference in the hardware cost ($2,000). 

So the countdown of servers started - 9,8,7,6,5,4,3,2..............1 server!!!!!
— Blast Off

The result -

1 physical server with 3 Virtual servers instances - that performs 3x faster than the previous system.   The way we setup the server makes it feel that the entire system in on SSD hard drives reading and writing at blazing speeds, with the added benefit of simplifying backups and management.

If you'd like our config and setup of this system we'll post it here - but it was just too technical to put in this post.   Thanks for reading - and thanks to our client for their trust in our solutions.

Thinking Fast and Slow - Workstations

Finally the new workstations have come in. These babies promise to increase the firms output exponentially, and they better for $4,500 per station. IT said that new computers were needed, evidenced by staff taking breaks during renders. The new machines took a month to get here due to spec’ing, budget negotiations and inventory limits, staff are past caring on IT promises.

But the workstations are finally here – aaaaaaaaaand then nothing. The Revit files actually appear to be running slower – even the laptops feels faster. So IT goes to work turning off layers and optimizing hardware performance, but it doesn’t feel any difference.

This is a black eye not only for the IT firm – but the Studio Managers reputation.

fast button.jpg
So what's going on?

Price tag does NOT correlate to performance.

 A workstation is not merely a box that does stuff, and Architecture software targets the system hardware differently. Examples below for software targets hardware

  • Revit – CPU speed and Ram
  • Adobe – Ram and Hard Drive speed
  • Rhino – Video card and Ram

The 2 companies with the greatest market share usage for Architects are HP and Dell. Both companies provide Architects with a decent machine. But they fail at understanding or conveying that a $1000 computer can perform better than a $5000 one based on what software will be used. The Architect through no fault of their own infers that “price correlates to performance” and sets an arbitrary price and hopes for the best.


If an Architecture firm has a great IT firm (or department) – they’ll understand where the money should be spent most effectively. But for the intro class here – get the second highest speed CPU with the most cores within budget. Everything else can be replaced or upgraded in 2 years to extend the life while prices come down.


For most Architecture firms an HP z440 configured correctly should last for 2 years without issues, then expect to spend $300 on upgrades to get another 2 years, finally being given to the Principal or Admin to last another 4 years. This rotation ensures efficient and practical use. Each trickle down additionally requires less IT involvement than providing new computers to users with simpler requirements.

The headline and impetus for this article comes from the book Thinking Fast and Slow: by Daniel Kaheman

Salient points from the book include:

  • When you can and cannot trust intuition
  • When experts are wrong - but everyone still believes them
  • Anchoring – setting a value on something before research


High Level bullet points

  • Price does not always correlate with performance
  • Focus on the CPU speed and cores – everything else can be upgraded
  • Explore the hardware needs and affinity of your software
  • Don’t extend the warranty. The cumulative cost is more than replacement or new computer
  • Workstation purchases are often non-refundable so benchmark before you buy


Want to race? Bring your files (happy to sign an NDA if needed) to our lab at 149 Madison Ave and we’ll show you how components affect software – schedule here. You can also join one of our free weekly AMA (Ask Me Anything – Tech Conference Calls ) to receive targeted expert advice regarding how to get your harddware and software to work well together.