Confessions of a Public GIS Manager: Does IT Outsourcing Really Save Money?

In following up on my example of a simple GIS app for entering and displaying lat/lon coordinates from a spreadsheet (or text file), the discussion went from cloud to client and then back to cloud. The reason may surprise you. Recall that I was looking for the best solution for a reader who was looking for a simple GIS app to display gobs lat/lon coordinates.

My first inclination was to use an online app (cloud) such as arcgis.com or Google Earth in order to stay away from the users needing to install and maintain software on their local desktop computers. No go. The functionality just wasn’t there. All along, my backup plan was to use a client app like ArcGIS Explorer. Well, after messing around a little and consulting with an online discussion group, that’s the route I went. I wrote about it last week.

Subsequently, a GIS manager from a public department (state level), wrote about his experience with client-based apps and his challenges with IT outsourcing. It really make one reconsider the cost effectiveness and efficiency of IT outsourcing. His perspective makes interesting reading:

We didn’t go the ArcGIS Explorer route primarily because of the current war (GIS vs. IT) which scientific computing is losing badly at this point in time. Our State and many others are neck deep in smelly muck created by business computing’s IT consolidation and outsourcing.  I just got back from a meeting where I heard another round of horror stories from VA.

For more than five years, our WAN-based users at regional offices throughout the state have ran GIS via Citrix with customized ArcGIS desktop apps written entirely in-house by our staff in VB and .Net.  We elected that strategy because at that time it was really the only serious option to allow access to the large amounts of data we had in our geospatial archive here at our headquarters. It was also attractive because back then we controlled our infrastructure and our LAN and were highly influential in WAN decisions because we had a very advanced computing environment here.

Then came IT consolidation and the predictable downgrading of advanced Agency’s capacity so we’d be able to open really big word processing documents on our desktops. By that, I mean scientific computing like GIS, Remote Sensing, etc. apps were not considered seriously in that process even though we need to operate much closer to DoD or NGA-like computing capacity compared to the average accountant. After multiple attempts to modernize our Citrix and SAN, resources were turned down and we decided we’d better switch to a new approach.

Because we’re charged serious bucks every time we put in a service call to have ArcGIS Explorer installed on an existing or new PC, we elected to go as thin client as possible. Everyone has a browser and we don’t have to pay to have that installed.  We initially developed some betas using ESRI’s JavaScript solution but browser differences (both different versions of the same browser and Microsoft vs. Firefox on individual PCs just inside our unit) caused many applications development problems so we abandoned development with that API.

That’s when we elected to do a very serious Flex vs. SilverLight comparison and the rest is history. The new beta has a rainfall widget we’re particularly proud of. It grew out of our active mining program staff having to respond to horrific flash flooding typical in spring and summer in our state. This new app will allow staff to go to the permitted sites to check stability of sediment control structures where the most rainfall (… based on Nexrad) was projected to have fallen for the first time this spring.

In April this year we’ll find out if our jobs are going to be outsourced or whether our state will modernize internally. The refusal to allow Citrix and SAN improvements is a harbinger of the way that will go I believe. We have been presented with 4 SAGs in the last decade. I wonder what the total count will after the first decade of outsourcing?

Many potential problems exist for geospatial programs because of IT consolidation and the more recent potential of outsourcing GIS. IT consol first. My unit does a great many very large (… and long) computing jobs. We routiinely move data from one projection or datum to another. When you deal with thousands of raster tiles, a reprojection project can take weeks to accomplish successfully. We also do spatial analysis projects that take even longer. We recently used Landsat scenes and higher resolution commercial satellite data plus aerials from multiple dates to do change analysis. That job took more than a month on beastly PCs we’ve built up specifically for these very tough jobs.

[[Our]] ICI is an old DoD concept I pulled back into use. We built these platforms after about a year of total frustration having our big jobs crashed from IT pushes of OS upgrades happening in the middle of producing badly needed new deliverables, network disconnects dropping out our license checkout connectivity to a remote license manager on the WAN, etc. I’ve already mentioned the failure to consider geospatial in upgrading infrastructure and improving bandwidth.

Even keeping your servers local can be a big battle in the war.  We have an older county size LiDAR dataset (pre-.las) processed and delivered as a point cloud. We have new LiDAR from the same county and we’re trying to do a comparison of the two datasets. Depending on what USGS quadrangles are selected it typically takes 30 to 40 minutes to load up four 1:24K quad size tiles of the older point cloud data via our LAN at fast Ethernet speed. Move that to a WAN situation and we need to start it loading in the evening so it’s ready by the following morning (but that won’t work because of the auto-shutdown software on all the desktops that executes every evening a 7PM). And then there’s the joke about the virus checking software pushed out to every desktop, configured all the same for everybody and auto executed and the twenty-one staff that mapped over five terrabytes of GIS datasets on the SAN and their very fast new computers (sarcasm) being brought to the approximate speed of molasses running up hill because the virus checking code never stopped trying to check all five TBs on each of the twenty-one PCs.  It wasn’t much of a joke when the whole Agency’s networking speed dropped to a crawl! Need I say more about one-size-fits-all IT mentality shooting off their own feet!

Negative aspects of outsourcing geospatial jobs are obvious. No contractor is going to know the individual program requirements like in-house staff and that’s a challenge even for us. Good example … the rainfall widget on the new beta app I pointed you to wasn’t requested by our mining folks until we approached them with the idea that we might be able to do something like that. Would there have been that kind of insights by a big corporate consulting firm like IBM or HP? I think not.

On the good side of IT consolidation, if geospatial folks are pulled together into a core group I think that gives folks the chance to work on a broader spectrum of tasks not limited by the bounds of what one state government Agency desires, but rather the state as a whole. That could be a good thing. Also, it gives Agencies with a GIS effort, consisting of one or two folks, access to experts they’d not be able to otherwise tap into (GIS DBAs, geospatial applications development gurus, etc.) and that definitely would be a very good thing. Of course that hasn’t happened here. On the good side, with outsourcing GIS jobs, I’m clueless as to how that could ever benefit anyone except the recipient of the contract. The horrible stories from colleagues give me night terrors.  PC refresh cycles of 5 years, horrendously expensive SAN storage rates, etc. You name it, &
nbsp;the customer is hosed by it.

There really is a business vs. scientific computing all-out war going on all around us and as I said initially, too many scientific computing types have their heads down doing the exciting stuff while the fight rages on, without them even knowing about it. If you can wake them up to the reality that business computing “experts” may very well be building a scientific computingless future in which they’ll have no place (or job), it would be greatly appreciated.

I’ll be writing more about this. It’s a serious issue and it’s not going away, especially with the geospatial industry continuing to put up strong growth numbers.

Thanks, and see you next week.

Follow me on Twitter at http://twitter.com/GPSGIS_Eric