NO. 03  VOLUME 01 - October 31, 2003



Sorry to be offline for so long.  Since our last issue, we planned and relocated 17,000 systems for a major telecom firm and worked on two World Trade Center projects.  We need to take a breather and catch up a bit on our newsletter.  


We were retained to advise an insurance company on the reality of what one firm did as a result of the WTC incident.  This very large company was not in the towers but was in nearby buildings.  Many of their windows were blown out and their computer rooms contaminated.  The Feds shut them down immediately.  One part of the story is how we were retained.  We were asked what we would do if we were the CIO of this large organization.  We advised the representative that we would go out and buy the latest and greatest of everything and send you the bill.  There was a pause on the phone and the response was "that's exactly what they did".  The interesting part of the story is that the new equipment replaced 11 platform levels of Windows and DOS.  Yes DOS.  They still had systems running DOS 3.1 and early LANs that were 3270 coax connected.  Now they have the latest Windows NT.  Amazing.  Our job was to assist in determining what was fair and reasonable on the insurance claim.  

Two cute stories.  When we were interviewed for the project, we were asked what we would do if we immediately had to replace everything in our large data center.  We replied that we would go out and buy the latest and greatest of everything and send you the bill.  After a short pause, the caller said "that's exactly what they did".  We were hired.  Second (sometime during the project), the policy holder had a large legacy 3270 environment that they replaced with an all IP-based network.  First question to us was "Is 3270 still available and if yes, why can't we just replace the 3270.  Our answer:  "If we were the CEO of a major corporation and our CIO said that they were going to replace their 3270 network with a new 3270 network to please the insurance company, two things would happen.  First, I would fire the CIO for being stupid.  Second, he would be the laughing stock of the computer industry.  We recommended that the IP-based network was a fair and reasonable replacement.  


We have a new utility on our website that will calculate your power and cooling requirements after you answer two questions: (1) how many watts per square foot in your electrical design, and (2) how large is the computer room (in square feet).  Press the button and you have information to immediately provide to your architects and engineers.  

Question:  How do I determine watts per square foot.  Answer:  Before you get to the two questions, there is a short 1-1/2 page tutorial on watts per square foot with several examples of legacy data center and server rack configurations along the watts per square foot for these configurations.  From the tutorial, you should be able to select the best watts per square foot number for your needs.  The formulas are based on known calculations.  To use this utility, go to www.abrconsulting.com and scan down the left side to the first link under the FACILITIES heading.    


Early in the design process, the electrical engineers will ask you to identify the power ratings for every piece of equipment that you intend to place in your new computer room.  If you cannot do this, the engineers will simply assign a watts per square foot to the design and move on without you (see Don't Be Left in the Dust on our website).  This also ties into the above paragraph.  What IT folks normally do is document all of the systems in the computer room and for power ratings, take the information off of the equipment nameplate (for example, a server nameplate that says 4.5 amps @ 120-208 VAC).  All of these power figures are then totaled, projections for growth in the new room are added in and the information is conveyed to the electrical engineers.  The problem here is that "nameplate" information is about 25%-30% higher than the design load needs to be because the manufacturers build such a large safety factor into their design.  Conversely, you have the running load which is the measurement taken while the computer room is in full operations.  Typically, the running load is 20%-25% of the design load (and sometimes lower).  

Here's what's going on.  The electrical engineer is looking for a design load for all equipment going into the new computer room.  This includes what could be 3-5 years growth to completely fill the room.  This design load then dictates the sizing of the UPS systems, the HVAC systems and the emergency generator.  Thus, supplying "nameplate" data with the 25%-30% safety factor is going to cause a large over design of your electrical and mechanical systems.  So, do you provide your engineers with the design load or running load?  Depends.  The safest spot is the design load less 25%.  However, if you have a UPS and a generator, you can lean more toward the running load since, the load is never expected to drop and re-start.  You can spend less on a UPS and generator.  However, if you only have a UPS, stay safe.  Your major decision at this point is how much run time you have on the battery should you have a loss of power.  You want to err on the high side when your only backup is a UPS.  


You can no longer avoid the increasing switch to digital telephone systems (Voice-Over-IP).  We see more and more being considered and installed.  It's a trend that will never reverse back to analog.  We have been designing cabling systems for the eventuality of VoIP for about 3 years.  Here's a summary of our basic design changes:

1)   All voice and data station cables are terminated onto the data racks (rack-mounted RJ-45 patch panels).  Voice station cables are no longer terminated on the backboard.  With the data switch and now VoIP switch being rack-mounted, this permits any of the station cables to be used for voice or data.  

2)   The cabling rooms themselves are more rectangular to accommodate more racks.  More racks are necessary to land all of the station cables (remember the voice cables used to land on the backboard).  

3)  Where VoIP systems will be installed, we now minimize the riser backbone copper pairs.  Digital telephone runs on the riser fiber pairs.  Where VoIP is in the future, we run the normal compliment of riser backbone copper pairs and terminate them on the backboard.  

4)   We then run 25-pair copper cables from the data racks to termination hardware adjacent to the riser backbone copper termination hardware.  This only has to be Cat 3 rated since it's just for voice.  On the data racks, you terminate these 25-pair cables on Cat 3 or Cat 5 patch panels.  Thus, when we cross-connect, all data cable patch to the data switch.  All voice cables are patched to the backbone patch panels on the data racks and then again at the riser termination area.  Once, VoIP is installed, this cabling system between the data racks and the backboard goes away.  

5)  All voice and data cables are rated no less that Cat 5 as VoIP systems will not run on anything less.  In today's buildings were are normally installing Cat 6 and occasionally, Cat 5e.  

Now, we need to get caught up with the UPS for the cabling rooms.  Most cabling rooms are not on the building UPS.  Guess what.  VoIP systems in the cabling rooms will go down with a loss of power.  This is a very foreign concept to users who are either on Centrex or a PBX with a 4-6 hour battery.  During a loss of power, you've always been able to pick up your phone and use it.  No longer.  When designing your new building or modifying your existing building, you must now include UPS power to your cabling rooms in your design specifications.  You must also now include the calculations for these systems into the projected load to be placed on the UPS.  

Finally, we suggest that you consider using the analog lines for your fax, Polycom and other analog devices for your emergency phones.  This means keeping them on copper circuits and not on the VoIP phone system.  The tradeoffs are worth it.  Once you see how VoIP works, there's no turning back.     


As we finish this newsletter, we see the announcement that Avaya has sold their Connectivity Solutions Group to Commscope.  This includes the Systimax cabling solutions.  We talked to our Avaya reps and they confirmed the sale.  They indicate that Commscope will retain them as a separate subsidiary.  Our opinion is "wishful thinking".  We've known that ever since Avaya split off from Lucent, the cabling business has been nothing but a stepchild.  Still, our Avaya reps have been wonderful and responsive and we hope that this doesn't change.  Stay tuned.   


Have you called Dell technical support lately?  How about Earthlink?  Called GE for a service call on your major appliances?  How about dozens of other firms with call centers?  Does everyone you talk to seem to have a foreign accent?  SURPRISE!!.  You are talking to call centers in INDIA.  That's right, India.  We were recently in contact with Dell on a software problem we were having with one of our systems.  After talking to about 6 different people, some of whom we could barely understand, we asked, "where is this call center located?".  The answer: "Bangalore, India".  "How many people work in your call center?" we asked,  "2,000" was the reply.  GULP.  The Dell representatives that we spoke to were very kind and patient but they were not able as proficient in solving our problems as the Dell people we used to speak with in Texas.  

Manufacturers and service firms are increasingly trading American jobs for $2hr-$3hr jobs in India - PERMANENTLY.  Reportedly, in addition to service calls, GE has outsourced the work done by 300 attorneys to $10/hr labor resources in India.  EDS reportedly will outsource 20,000 programmer jobs shortly to India.  $2/hr. vs. $58,000/yr. is the justification.  

Recent articles indicate that the U.S. is losing hundreds of thousands of permanent jobs to India.  It seems that once we can commoditize a service, we can ship it offshore.  It seemed to start in 1998 with Year 2000 programming fixes where an estimated 25% of all Y2K work was done off shore.  Since then, data centers have been built in large Indian cities.  These cities themselves experience unstable power, shortages of water and other deficiencies.  But, not the large data centers.  They have been built to be failsafe.  These data centers are big money to the local economies and the people are highly educated, highly motivated and will work for a lot less than you can.  Large call centers are also appearing in Nova Scotia and in Calgary, Alberta, Canada.  Expect more outsourcing to these less expensive areas.    


As a result of being brought onto projects after the budgeting is complete and the design has begun, we advise you not to overlook some very costly items that you did not plan to see as part of your renovation project.  We're talking about having to bring certain types of systems up to more current codes as a result of pulling permits on planned work.  For example, one firm was upgrading their UPS and adding additional wall panels for more equipment.  They were required to bring all electrical systems up to 1998 NEC code.  Ouch!  On another project, a major telecom local provider has 5,100 pairs in a demarc serving two 17-story buildings.  The cables were installed three decades ago and are not on fused protector blocks.  We raised the issue with the local Fire Marshall and he said that they had to be brought up to NEC 1998.  Hello fused protector blocks.  Now the building owner and the carrier are in discussions to see who pays for the upgrade.  You would think the local carrier would be fully responsible.  They don't think so.  Still another example is a new client that is upgrading their computer room.  The room won't expand but their will be significant modifications to the room.  They have an older Halon total flooding system in the room.  Hello $60,000 for an unplanned upgrade to FM-200.  Hopefully, by reading this article, you can investigate those areas that you think may be grand-fathered in a renovation.    


A problem that we continually encounter on new construction projects, and a problem that we dealt with on one of the projects above, is the date or dates that you will have access to your IT spaces for what is called "customer fit-up".  This includes access to the computer room, MDF, the TR's and possibly specialized labs and other IT spaces to install racks, cabling, routers, switches, PBX equipment, etc.  The problem occurs when your well planned 30-45 day for fit-up of the initial routers, servers and PBX prior to the beginning of the move events erodes to about 2 weeks (and sometimes less).  In our project, there was an important change in the electrical design that caused a delay in delivering the buildings on time.  Facilities would not push out the move dates so IT went into a "heightened state of activity".  We made it on time but it was a lot of work.  

What causes the problem?  The largest part of the problem is not being directly involved in the construction meetings where timing and scheduling is discussed.  Second, the general contractor's main objective is to deliver a complete package and not just the IT spaces.  The GC realizes your needs and understands your request but delivering your IT spaces while other major parts of the building are not yet complete is in general, not the way buildings are built -especially large buildings.  So you have to be diligent and persuasive.  

So how do you get smart about this potential problem?  First, obtain a copy of the construction plan well before construction begins and review the dates.  Make sure that you can determine from the dates when the IT spaces will be turned over to you.  Often times, it is the GC's expectation that you can't accept any premises or begin fit-up work before he's done.  It that's your expectation, everything is fine.  However, if it's not, this discrepancy needs to be resolved while the steel is going up and not closer to the end of the project.  Second, locate the dates for the paint, carpet and linoleum start and finish dates.  Next, understand that these dates are often "place holders" and may not reflect the actual dates.  Understanding these dates will give you a good idea of if, or if not, you can expect your IT spaces on time.  You also have to follow the notes of the construction meetings to make sure these dates firm up.    

Note that you can begin "Customer Fit-Up" before the Certificate of Occupancy (COO).  This includes the installation of carrier lines, core routers, switches in the cabling rooms, the PBX system and initial servers.  It is reasonable to get your new facility up and on the company network.  What you can't do prior to the COO is to go into production.  Usually, this is by prior arrangement with the GC.  You may have to do a punch list of of the IT facility being turned over and assume all responsibilities for the room once you have it but it can be done.  What the GC doesn't want is somebody in his rooms that can scratch and damage the room before it's turned over.  


From time to time, the ABR Consulting Group, Inc. will e-mail you condensed information for the ever-changing and every-dynamic world of designing, planning and relocating data centers, server rooms, labs and other key IT/IS spaces.  All you have to do is place your e-mail address below and click on subscribe.  You can unsubscribe at any time by entering your e-mail address and clicking on remove.

Email Address:

Contact us at www.abrconsulting.com  
Phone:  925.872.5523