Healthcare IT Blog

It’s the Network–Still.

Published on 11/06/2012 by Joe Kelly
Category: Networks

I’ve been involved with technology consulting in the MEDITECH space for a solid 15 years.   Prior to that, I served as either a Technology Consultant or Engineer for an additional 18 years in a variety of industries ranging from hi-tech to high finance.  Most of that time has been spent one way or another on planning, designing, assessing, and implementing networks.    All kinds of networks:  voice, data, video, analog, digital, wired or wireless.   I’ve touched it all.   I was around when it was an analog world.  Heck, I did all my college programming using a 300 bps acoustic coupler.   I was there for the conversions to digital and fiber.  I‘ve been around long enough to remember when T1s were a big deal.   I used to install 2400 bps modems that were about 25 pounds and bigger than a desktop PC.  Today I’m planning converged networks using 10 Gbps fiber optic-based networks.  I was reflecting on how long I’ve been doing this the other day, and I came to two inevitable conclusions.  The first is; WOW, I’ve been doing this a really long time.  The second is; the more things change, the more they stay the same.

I have an example that I use when I talk about planning and networks.  It’s especially effective in a group setting.  Typically, I’ll draw a circle in the center of the whiteboard and label it “Network”.  From there I’ll draw various systems.  I’ll ask the room what systems are currently deployed.  Inevitably we’ll get MEDITECH and PACS and MS Exchange.   From there I’ll get and all kinds of other systems.   When I’m done identifying systems, the picture looks like a huge wagon wheel with dozens of spokes.  The next question I ask is, “Which is the most important computer in your Information Technology environment?”  There’s usually a debate about which is most important.  Everyone agrees that MEDITECH is pretty important.  There is, however, debate about which is MOST important.  Oh sure, some folks will identify PACS as being pretty important too, and some may say the standalone billing system should be considered.  Some folks will argue that the UNIVERSE or Active Directory servers are most important.  I’ve even had folks argue that email is most important.  But agreement is usually elusive.  After a minute or two of debate, I weigh in.  Just about the time it’s clear that there is no one most important computer, I draw a big circle around the network.

Think about it, if the Active Directory server does down, you can still get to LINUX apps.  You can even get to MEDITECH (depending of course on which version of MEDITECH you are using).  If MEDITECH goes down, you can still send emails and get to PACS.  However, if the network goes down, no one gets to anything.

This is as true today as it was thirty-three years ago.  Only today, the exposure is greater.  Think about the environments we have to support.  It used to be you had a dumb terminal plugged into a cluster controller connected to a front end processor.    Controllers connected to the FEPs over relatively low speed networks, typically 9600 bps.  Today we have virtual servers and virtual storage and virtual desktops.   We need to support fat clients, thin clients and zero clients.   Clinicians are using laptops or desktops, or iPads or Smartphones, and just about anything else you can think of.   Have you seen the new TVs that are Internet enabled?  Will we be supporting access to clinical information from TVs next? The answer, of course, is YES!  The infrastructure this is all supported over is wired and wireless.  The wireless network can be Wi-Fi or broadband.   Network speeds run from 10 Mbs up to 10 or even 100 Gbs.  And the infrastructures are all becoming converged, to boot.  We put a Converged Network Adapter in a VMware Host and all of a sudden multiple virtual servers are accessing storage (Fibre Channel) and Ethernet networks over the same physical cable.   Eventually it all ends up in a network switch somewhere in the data center.   (The mind reels.)  Now consider the possibility of that little (or big) top-of-rack switch with the single power supply fails, and *POOF*, it’s all gone.  Nobody gets to anything.  Unless you’ve designed resiliency into your network.  Then the question becomes how much resiliency, and to what level?

The problem has been the same over the multiple generations of technologies that we’ve run through in the past three decades.  It will be the same in the three decades to come.  All of the spokes in our Information Technology wagon wheel meet at the hub.  In this case, the hub is the network.   No matter how you spin it, everything touches the network.  When designing infrastructures, especially virtual ones, it makes sense to consider the impact of and on the network.   I can’t tell you how many configurations I’ve seen that include dual-attached, load balancing network configurations where both network connections terminate to the same physical network switch.  Typically, when asked why, the response is: “We couldn’t afford two network switches”.   I think the question is: can you afford not to have more than one switch?

I’ve read and seen dozens of consultants talk about the cost of downtime.  The cost typically ranges anywhere from a low of $265.00 to over $1,500.00 per minute for a 500 bed hospital.  Simply put, the cost of downtime = (outage duration x 2.15) x downtime cost.  Assuming an average cost of $1,000.00 a minute (the math is easier), the cost of a 1-hour network outage is approximately $129,000.00.   Think about it, 99.99% availability results in about an hour of downtime per year.  Considering the far-reaching impact of a network outage, whether physical or virtual, when designing a technology infrastructure it makes sense to start with the network.

It never changes.  It’s always the network.  The technology changes but the impact stays the same.  Servers and storage get all the attention, but the network carries the load.

Joseph Kelly is the Director of Technical Consulting at Park Place International.  Joe has been working with Healthcare providers and/or payers since the mid 1980’s and focused solely on MEDITECH and MEDITECH hospitals since 1997.  Joe has provided technology consulting, architecture, design and planning services while at organizations such as EDS, JJWILD, Perot Systems, Dell Services, and now Park Place International.  Joe’s overall all goal is to bring the leveraged, cloud-based virtual universe down to earth to most effectively meet real world objectives for MEDITECH hospitals.  Joe has a BS in Computer Information Systems from Bentley University.