Higher Education

Hello! I'm hoping that some of you  are or have been involved in deploying 100% wireless across your campus. What I mean by this is both a.) complete coverage across all your buildings, and b.) migrating the vast majority (if not all) clients onto wireless. For those of you that have done this or are exploring the same, can you take a few minutes to share your experience via the following questions?

- What was your motivation for pursuing 100%?
- What methodologies did you use to estimate APs required for the project for budgetary purposes? (e.g., square footage, client-to-AP density, ...?)
- What are some of the hurdles you have encountered with this project?
- As you deployed, were any changes to design/approach necessary?
- Similarly, how accurate was your initial estimate of how many APs would be required?
- Have you completed 100% deployment?
- Was outdoor included? If so, how did you approach estimating/budgeting for outdoor spaces?
- What percentage (hunches/guesses acceptable) of client devices were incapable of moving to wireless? Why?

I really appreciate any intel you can share, as we'd love to have other comparables as we go down this path. I imagine answers may also create some interesting dialog between schools.

 

Thank you!

Hi Ryan,

 

Are you looking for comments targetted to academia, or would you find thoughts helpful more broadly speaking?

 

I.e. I've worked with roughly 20 large scale academia bodies (student count higher than 20k) in the UK, and more than 100 "small" ones (less than that many students obviously). This has varied from Unis to what you call elementary schools. Beyond that, I've worked with campus deployments in hospitals and large enterprises.

 

So, would you like me to limit my input to academia related, or open it up a bit?

Great question. I was largely focused on academia due to the likelihood of there being significant mixtures of building types, from newly renovated ones to older, historical buildings with more difficult RF challenges. The other thing that comes along with academia is the heterogenous clientele, which is fundamentally more challenging than a typical enterprise where the majority of devices are controlled and managed by the business.

The large scale bodies you mentioned are definitely of interest, however, if you find value in any other locations, yes, please do share!

Thanks!

Hello Ryan,

 

With your comment in mind...

 

My point of view is somewhat unique, in that my buisness is a vendor-agnostic professional services setup. I.e. we do everything (wired and wireless) on all "tier 1" (and most tier 2-3) vendors in the IT channel (indirect service delivery). As a result, our views tend to be more blended in some respects.

 

So (and sorry if this is a bit long, you might wish you hadn't asked)...

 

- What was your motivation for pursuing 100%?
Most customers I've worked with pursue this to achieve Wi-Fi VOIP/IPT or RFID outcomes. Hospitals and IPT heavy enterprises fit into this slot most often. Others that target 100%, often do so as a result of lack of governance or assurdity from within the buisness regarding "who will be doing what, where and when". Those without Wi-Fi use cases mandating 100% specifically, tend to target key coverage areas (and this represents most of academia). I trust this makes sense?

 

- What methodologies did you use to estimate APs required for the project for budgetary purposes? (e.g., square footage, client-to-AP density, ...?)
The short answer from me, is this is a cost vs risk decision. We deliver multiple "packaged" services into the IT channel that deliver against this requirement, and there's no single answer. We seek to work with the reseller/distributor/customer in order to understand where the project lifecycle is, and what the world looks like from a cost-accuracy vs. budget-for-planning point of view. What we deliver, is what the requester asks for, and we ensure they know what they are getting from us including associated risks from any service. Whilst in the technical world we seek to de-risk, the commercial world doesn't always support that of course! For all services in this space, we have a PQQ (pre-qualifying questionaire) to be filled in by the customer which includes questions around predicted use cases for the Wi-Fi, anticipated client count etc (all in "simple" non-techy language). Examples of "how" we go to work are as follows...
a. If the view is nobody has any money for planning, we have a spreadsheet that calculates an amount of APs based on square footage (adjustable based on Wi-Fi use cases and user count IF known (which it usually isn't)). This is normally requested by "lower skill, higer risk" partners who are simply speculating against opportunities with customers. We of course advise the partner we are not responsible for any degree of accuracy, and this is simply a well intended guess provided "as is". Accuracy tends to be around 85%. On occasion we've extended this to "pen-on-paper" in special circumstances.
b. Offsite planning. This is what it sounds like. RF modelled in Visual RF, Airmagnet planner or Ekahau. We tend to use the tool that most closely aligns to the partners vendor strategy (e.g. Visual RF for Aruba). This is a chargable exercise (as we tend to spend time modelling it properly with walls etc). Again we ensure that partner knows we're not accountable for the accuracy gap. Why? Because I can't protect the business against onsite "unknowns". I can't demonstrate my assurance of accuracy with FACTS. This service is usually used for partners who want colateral that is more compelling as part of bid/proposal/budget, that also sends a nice message to the customer (pretty pictures). Accuracy tends to be around 90-95% (as all my engineers who do this, have to have a number of "real" surveys under their belt). If you look at example from those that haven't, it obvious to those that have, right?
c. Onsite planning. This is proper surveying. AP on stick with Airmagnet. The only gap of course being "if the site conditions change after we leave, results will vary..." which we cover off. We do underwrite our work with this service, so the customer is de-risked. This service is used by risk-averse partners who want to establish full cost accuracy. Broadly speaking, it's commonly used by partners who are Cisco (about 70% of our survey work), and to a lesser extent Aruba/HP/Motorola based. Partners who sell (let's call it) "cheap stuff" just don't get it, and can't rationalise the cost for the work.

- What are some of the hurdles you have encountered with this project?
If I focus this answer against 100% scenarios, it's usually AP fitting in challenging areas. RF maximisation vs. practicallity sort of thing. Again, the simple answer I'm afraid is it's a cost decision. No matter how great the desire for 100%, no sensible business will spend x10 cost to install an AP on a newly erected mast in a field (with power). Architects can be a bit precious. They don't like APs on shiney new ceilings. I usually leave this battle between the prime building contractor/architect and head of IT (I have seen them go to CEO level), and the outcome is 50/50 between "APs above ceiling" (RF impact) and "APs below ceiling, or special antennas". Usually, biggest personality wins (rarely comes down to contract Ts&Cs). I see asbestos has already been mentioned. We see that quite a bit in the UK. Stairwells can be an interesting one too. Frequently cablers struggle more in those areas due to wall thickness and firebreaks. So costs go up, and get challenged. And lastly of course, APs getting stolen in open areas with low ceilings is something to consider.

- As you deployed, were any changes to design/approach necessary?
I think this has been covered nicely by the previous comments (Kevin). Note that as an interesting blended metric, we see CAD diagrams to be inaccurate 80% of the time. However, the inaccuracy only tends to impact planning about 10% of the time. As long as outer walls and heavy inner walls are there, you're fine. You more or less always get correct plans for new buildings. Where the plans are inaccurate, we just ensure the customer understands why an AP is shown on the plans in "strange places".

- Similarly, how accurate was your initial estimate of how many APs would be required?
With approach "a" above, we hit about 85%. For approach "b" it's about 95%. For "c", it's 100% obviously. But of course I use RF skilled guys to do this. If it's done by people who've never touched Airmagnet in the wild, your accuracy goes right down. You can't teach it in a classroom.

 

- Have you completed 100% deployment?
Broadly speaking, those that go for it tend to fall in early budget stages, or they do get to the goal yes (40-60 split). Having said that, I have seen metric evidence on occasion that suggests ROI in certain places is quite low after completion. I.e. APs in certain places never get an association.

 

- Was outdoor included? If so, how did you approach estimating/budgeting for outdoor spaces?
Up to about 2 years ago, virtually nobody wanted this. Now, the shift is towards including it. Over the last year, 75% plan to include it. About half of that actually executes against outdoor (again, cost based).

 

- What percentage (hunches/guesses acceptable) of client devices were incapable of moving to wireless? Why?
This is rarer these days, but you still see it. Track back 3-4 years, it was around 15%. Now, it's less than 5%. Wi-Fi is a good catalyst for ditching legacy kit. Mostly, this comes down to third-party bespoke build devices that are unsupportable (but still in use). Interestingly, 3 times in the last year, I've used Wi-Fi considerations/requirements as catalysts to help the justify removal of XP (and excelerate associated projects). Rationale varies from security requirements, GPO practicality, NIC age etc. Customers like this once they come around to the idea.

 

Hope this helps.

Ryan-

 

I can speak to the (nearly) 100% campus coverage piece but there never was a unified effort to eliminate wired clients here at HWS.

 

- What was your motivation for pursuing 100%?

 For us, it was a confluence of two issues.

Firstly, our previous campus wireless network was planned in stages over a few years.... right in the middle of that was the economic fallout of roughtly 2008, so our previous wireless network was never fully implemented, and what was there was reaching EOL quickly.

Secondly, and the more substantial contributor to us moving forward was the outcome of a cohesive student satisfaction survery.  Wireless (or lack thereof) bubbled up as the top issue for students.

- What methodologies did you use to estimate APs required for the project for budgetary purposes? (e.g., square footage, client-to-AP density, ...?)

This was an enormous challenge for us.  We were fortunate to be in a place where we could plan for a full blown rip-and-replace and we went through a robust and diligent RFP process.  It was during the evaluation phase of each vendor when we were refining the estimates where some vendors shined, and others failed.

We are a campus of 170 acres with 100 or so buildings.  We are not a college with 5 or 6 duplicates of buildings either (like a bunch of dorms built all at once), so save one or two unique cases, each building would need an independent plan and had unique layouts and building constructions.

I will say this was an iterative process.  Early phases included -  We did our best to categorize buildings by age/size/construction method, let vendors plan (however they wanted) for one representative of each, and did some extrapolation as to how many APs that *might* scale up to.  We definitely asked the vendors to participate at this level... we had at least one player doing interesting antenna stuff with some interesting claims as to coverage, and another vendor whom we had some questions about their capabilities in this arena - the results from this phase were very telling.

Beyond that Aruba had been good enough to take CADs of *all* of our buildings and do a quick auto-placement in Visual RF for all buildings.  We knew this wasn't perfect and would require refinement; but it got us in the ballpark.

We then did a pilot/proof-of-concept in one or two buildings to prove coverage predictions *and* try on some wiring contractors for size (and start to build out wiring estimates for AP installations at the same time).

At this point we were comfortable moving forward with Aruba and did some iterative refinement (together) of the Visual RF plans, combined with some site surveys in key buildings and nailed down our best guess.

In the end, despite *moving* a lot of APs from the original plan, we ended up with a negligble delta from deployed vs estimated APs.  I will say we are probably a bit on the high side of AP-density but we were conservatively planning - and I'd *much* rather deal with too dense than not dense enough.  It all started with AP counts - everything hinged off of that - licesning; wiring/installation costs; POE ports...

- What are some of the hurdles you have encountered with this project?

Your previous question nailed it.  We knew we were headed into a large, expensive project - but we didn't know how big the breadbasket was going to be. 

Also Asbestos.  We didn't have a complete picture from our B&G folks as to where and when we might run into it.  Large building count with lots of variety in construction style and a short installation window.  In the end we only fought it in one building; *but* it was the elephant in the room throughout the installation - and we moved exceptionally fast in deployments with the thought that we might need some unplanned extra time on the back end of the timeline if we had to deal with it.  So in some sense it was not so much of a problem as much as it was a risk/fear that guided our timing and decision making along the way.

- As you deployed, were any changes to design/approach necessary?

We did make slight changes as APs were installed.  If a wiring contractor was going to have trouble (or couldn't) installing an AP per the design, there were times it affected other AP placements. (Also occured when the CADs we had were not accurate)  This was an on-the-fly experience as well as we were installing something like 800 access points in about 6 weeks in about 80 buildings.  A little hair-on-fire for sure, but we got it done.  It helped setting some ground rules with the contractors (like "you have to come get us if you can't get it into a room" or "in a smaller house, if there is only one AP on the floor and you need to go one room over, that's ok - otherwise come get us) and the contractors for the most part were good about finding pathway problems before they got too far into a building.

I will say also that it was the smaller houses that the pathways and AP placement adjustment occured the most - and fortunately - these were the locations most tolerant of adjustments to the rf plan.

- Similarly, how accurate was your initial estimate of how many APs would be required?

See above.

 

- Have you completed 100% deployment?

For the most part.  There are definitely "middles of parking lots" that are not reached and locations we knew we were not going to cover - but otherwise we've got full coverage.  I will say again though - we were fortunate to be funded for a full blown rip-and-replace so it was a once-and-done installation for this generation of our Wireless network.

- Was outdoor included? If so, how did you approach estimating/budgeting for outdoor spaces?

Outdoor was included, for the marjority of notable quads/rec/fields/student locations.  We identified where we wanted coverage, and our local SE had an outdoor planning tool (predictive) which we validated and supplemented with surveys when they went up.  We have added a few outdoor APs as we hit one or two outdoor shadows but otherwise coverage has been good.

If I have one gripe it is that the rf planning tool for Outdoor APs and visual RF aren't integrated; and Visual RF doesn't account for building adjacencies.  Your indoor cells are naturally going to be smaller than your outdoor ones... but there is adjacency and has made some things interesting for us (as have closely adjacent buildings).  In a horseshoe configuration of three buildings, we had an outdoor AP in the middle to cover the 'quad'y area inside of it which I ultimately completely turned off.  We had enough bleed out of the buildings/students interior to the horseshoe were associating with the louder outdoor ap from inside the buidling; CCI, etc.  I am wondering now if some of the ARM 3.0 controls along with a max tx getting dialed back on the outdoor AP would fare better than previous AOS versions (now I'll have to dig...).

- What percentage (hunches/guesses acceptable) of client devices were incapable of moving to wireless? Why?

N/A.  Anectodally though, our student Res Switches are *dark* save the AP uplinks.  Still pulling numbers over time, but it is staggering how little they get used. 

I hope this helps, even if it is a touch long winded.  I know little old HWS is no OSU, but I'm happy to fill in any gaps/discuss our experience as it