20 December 2010

The Big Lie #2; “Public cloud services save organizations money.”

A few years ago, I rented a washer and dryer for a new apartment.  I paid $42 per month and used them for 18 months, ultimately paying more than the pair would have cost to purchase – in just 18 months! In a possibly more personal example for many readers, the US self-storage industry has 78 square miles of storage space generating revenues of ~$24 billion to store junk worth less than what’s ultimately paid in rent.
Let’s move over to communications and utter the moth-eaten word, “CENTREX.”   You remember, the all-knowing beneficent Telco arranged for you to pay a few pennies a month and voilĂ , the PBX-born headache magically disappeared.  Except a smarter competitor got more features, was more agile, took aspirin for the headache and ultimately owned the depreciated asset for years while the CENTREX continued to pay Ma Bell and in the end got nothing.
At what point would the CENTREX customer have owned the better-featured PBX?  Certainly in less time than the average PBX lifespan of 10-15 years. 
Let’s return to the present where everything old has been repackaged again. 
Before we begin, I want to differentiate the public cloud from a private cloud (or data center deployment).  I believe organizations should own assets and expertise, not rent them.  While I like the architecture that cloud deployments offer, they should be installed in an organization’s own data center.
Also remember that it’s a vendor’s job to pick your pocket.  Like a gold-digging bleach blond, vendors only like customers for their cash; no budget, no love.  And like any good gold-digger, there is always a stream of “events” where big money is needed.  Twenty years ago it was CTI in the contact center, then it was unified messaging, then multimedia contact center and most recently UC, collaboration and video.   On the positive side, given the number of “dialtone and voicemail” RFPs that cross my desk, most organizations are wise to their gold-diggers.
Personally, I believe that the two reasons multimedia failed in the call center came down to value for money and most call centers’ secret desire to train customers not to contact them, ever.  Think about it, 30 minutes on hold and a frustrating dialogue with Paul/Sanjeev is a much richer deterrent for a consumer than a quick text chat with an agent to solve a problem. 
Just today, during a visit to a large retailer, I enquired whether a perishable product was available at the branch closer to my home.  The clerk’s answer was that she didn’t know and didn’t have the phone number to the other branch to find out.  She returned to her Christmas carol induced fugue knowing I wouldn’t bother her again. 
But I digress…
Traditional Telcos are losing landlines faster than some men’s hairlines.  Supplying data and SIP truking just ticks them off; AT and freaking T is the architect of modern communications, it doesn’t “do” PLUMBING.    Adding “value” (read: services or gatekeeping the internet) enables them to save a little face and generate fatter margins.  Because, news flash, just like my washer/dryer, users always pay disproportionally more than they get (Hello, SMS messaging).
Vendors’ nifty diagrams and amortization plans all seem to show hosted solutions being more cost effective than a CPE system in the 5-7 year range (depreciation range).  However, elephant in the room, the average life expectancy of a PBX is between 10-15 years.  Hosting is predicated on the (I believe erroneous) belief that we have entered a new era where communications infrastructure, like PCs, will be replaced in under 5-7 years.  In addition, those payback calculations assume that a customer would be keeping up with every insignificant change over the life of the system.  We all know that’s pretty rare, don’t we, all you happy Definity customers?
Public cloud hosting, CaaS, SaaS or managed services for that matter is simply the creative accounting of shifting money between buckets which ultimately ends up costing users more.  Customers figure they don’t have that nasty PBX to maintain anymore so they can fire Joe, but the hosting provider needs Joe and charges the customer disproportionately more for Joe’s services.   There may be some comfort in receiving a largely unchanging regular bill for communications services, but bill smoothing isn’t savings.
There is also a lie within this lie called multi-tenancy.  The news today is filled with virtualization this and virtualization that, but all that means is that they can stuff the same server counts onto virtualized servers within a blade cage.  Or more simply put, servers are getting smaller and more applications can be run in the same physical space (blade) but each one is a separate instance of the application.  Think of it as duck-taping two laptops together.  They both run Windows and PowerPoint but they are walled gardens.  There may be a minor hardware savings, but that’s not where the big money is today.
True multi-tenancy would really save administration, management and updating time because it’s a single instance with a shared management portal controlling multiple companies’ systems.  Telcos do this all the time; a cell phone isn’t administered on a single dedicated box.  One portal to control many separate customers. 
But multi-tenancy is hard.  Traditional vendors don’t do it.  Cisco doesn’t; Unified Computing System (UCS) is one per customer. When Verizon or BT are seen offering solutions hosted on UCS, picture a locker room with a customer name on each locker. Interactive Intelligence is the same.  Avaya sounds like it can’t catch its breath, h-h-h-h-h-osting?  Alcatel-Lucent gives a French shrug. 
However, Broadsoft can do it. Certain start-ups can do it like LiveOps and InContact in contact center.
As usual, it’s up to end users to pierce the PowerPoint veil because vendors like to turn limitations into must-haves.  This explains why Cisco and Interactive Intelligence talk about the added security of having a dedicated box.  In Cisco’s case this might be understandable; after all, their June 2010 users’ conference (CiscoLive!) was hacked, compromising up to 20% of attendee records.

The other big reason not to go with public hosting is vendor lock-in.  Never underestimate the difficulty in repatriating and rebuilding the data given to the hoster during the initial setup and ongoing usage.   Most organizations will slowly move to hosting to see if it’s right, but even the slowest rollout may not be enough.  After all, the same forces that left customers with 7 different PBXs would likely see future regime changes wanting to leave their own imperator on their IT kingdom – the executive version of rearranging the furniture (and they ALL do it, don’t they rank and file?).
So, are there any reasons to embrace the public cloud?  The answer is a qualified, yes.
First, it’s critical that any organization seeking a hosted solution via a hosting service needs to know who the underlying technology is.  Are you buying Cisco UCS, Broadsoft or Genesys?  Each has their own plusses and minuses.
Quick setup and tear-down of users.  In a LiveOps whitepaper written by Sheila McGee-Smith, one case study talks about a user who has a relatively stable pool of agents during most of the year but experiences a huge swing during the end of the year.  This customer is an HR outsourcer and I suspect they are referring to the period of open enrollments that happen at the end of each year.  However this would also be the case in certain retail environments where Christmas causes a huge uptick in contact center needs.  While this makes perfect sense in some contact center scenarios, if an organization’s rank and file staffing needs swing this wildly, there’s a bigger problem afoot (unless you run a tax preparation service). You can read the full whitepaper here but be warned, LiveOps is one of those companies that views any casual interest as a prospect, so you must register to view it (of course you can make up fake information like I do).
Corporate indolence.  If an organization is slap-dash about keeping systems up to date or staffing up to snuff and has suffered negatively because of this laziness, by all means outsource the problem as a work-around to organizational limitations.  (Yes, Northern Telecom Meridian, Lucent Definity and Rolm 9751 customers, I’m talking to you.) 
Corporate frustration.  Another user in the above mentioned LiveOps whitepaper spoke of the myriad of systems collected over time either from supplier course changes or acquisitions and the difficulty in managing and updating the interconnected systems.  However their moving to the cloud changes the dynamic from too many vendors to, “is this the correct vendor” for the next XX-years?  (The same question had they united under a single CPE vendor).
Communications isn’t our core competency.  Unless you’re a small business with limited resources, this is just weak.  It’s the difference between knowing something and knowing how to Google the knowledge.  One offers direct control, the other is speculative. Can a librarian change a light bulb or must an electrician be called?  
But just as there are reasons to consider the cloud, there are reasons not to.
·         It will cost less.  Scroll up.
·         It’s more secure.  Ask Cisco or any other hacked organization.
·         It severely limits an organization’s ability to change.  Hosting is the bedmate of vendor lock. 
It’s my opinion that end users should PaaS on CaaS and SaaS unless there’s a very specific reason to do so.  Otherwise, I have a WaaSher and dryer I’d be happy to provide.