Monday, September 07, 2009

Cloud Computing: Still Bad.

In response to my previous post, Cloud Computing is Baaaad, Mmmkayyy?, I was happy to receive a most thoughtful comment from Greg Charland. Greg writes in part:
Really, though, it comes down to the math.

I'm not even going to go in the direction of google as I'm not convinced that their email solution is business-worthy.

But I can offer hosted Exchange & SharePoint through a well-known provider for $15 per mailbox per month...and make good margin doing it. Not cheap crap but a real provider who answers phones, monitors their servers, and knows their stuff.

For an "average" 10-user business:
10 x $15 = $150/mo
That's $1,800 per year
That's $7,200 over the 4-year life of a "typical" server.
I invite you to read the entire comment, please. I've been writing this blog for a long while, now and I'm thrilled to say that this is the most thoughtful, insightful comment I've ever received, well worth the read.

Nevertheless, it does not change my mind. My core advice to anyone is summed up in the statement "Do not put your mission-critical data in someone else's hands," (and I mean this generally, but more specifically I mean "in their hands alone"). With all due respect to Greg, that means don't entrust it entirely to Greg... or to ME for that matter. The rest of this post is addressed to Greg in response, but the rest of you feel free to read "over our shoulders."

Greg, cost savings isn't the whole of the equation, and I suspect you realize that... otherwise you'd have no alternative but to go in the direction of Google mail and docs, which can provide pretty much what you do, but for free. I realize that you're not convince that they are "business-worthy", but you yourself face precisely the same hurdle. I've been burned by "well-known" cloud vendors who talk a great game, only to shut down, taking my data with them. Your challenge is not only to convince someone that you're more "business-worthy" than Google, but that you're in it for the long haul.

As to costs, per user you are charging $15/mo x 12 months = $180 per seat. For about $140/seat I could set a business up with Domino Collaboration Express and Notes clients, and Lotus' small business pricing would effectively throw in the Domino server for free. The savings is applied toward the purchase of server hardware, which for 10 users doesn't need to be that hefty. Not only are email and file sharing services lightweight, but groupware tasks can be distributed to the workstations as appropriate. Most small businesses overspend on server hardware, and that's simply because they're told to do so by consultants who sell the hardware. The "typical" server cost of $10K is completely overpriced (for SMBs, more on that later), and that higher-than-necessary rate is commonly quoted not only because it's what small businesses are suckered into buying, but because it serves the interest of SaaS vendors to quote the higher price. For a small business server, the actual cost is about half of what you're quoting, assuming that their needs are reasonably constant over the life of the box.

My challenge would be to convince a client that this localized solution is worth the nominal additional expense. And it's not much of a challenge, really.

First of all, compared to Exchange, Notes/Domino is a an overall better solution, as detailed in this comparison on ChannelWeb.

Then there is the security aspect, which is a major factor. They would own the hardware, they would have 24/7 physical access to their data, and no one else would. Their internal network is not dependent on any external party, nor any periodic renewal, nor any terms of service. Their ability to do business is not hampered by the failure rate of their internet provider + that of their hosting service provider.

Next, there is the flexibility of the solution. You would offer Exchange and Sharepoint. A local Domino solution could also host additional services, as mail and groupware would not begin to tax its capabilities. At zero additional cost the business has available to it any of the features provided by the many Domino database templates that ship with the product, supplemented by those which are available at OpenNTF.org, and available from third parties.

The same hardware could host accounting or other shared services beyond those which the hosted service provides. Keep in mind that we're only discussing 10 users here... this is extremely lightweight. And here's where your math falls apart.

You must realize that if the user needs to provide these services which are not provided by your host, he's going to have to buy the hardware anyway. Then the equation changes drastically. Mail and groupware for 10 people are easily provided in the spare CPU cycles of the hardware he will have on hand, and the purchase price of this hardware with regard to mail and groupware is properly removed from the equation entirely, leaving the user with a $400 acquisition savings in the first year, plus savings on subsequent renewal fees, compared to hosting. The yearly savings defrays the cost of on-site maintenance and support (which, again, would be paid anyway), leaving the small business with all of the security benefits discussed PLUS "free" email and groupware after the first year as compared to your solution. So, after the first year, and for the limited set of services we're discussing, the local solution is pure savings. This doesn't even factor in the benefits of additional template customizations or custom groupware solutions not provided by the host.

As you say, it comes down to the math, but cloud computing doesn't look one bit better when you REALLY look at the numbers.

Now, Greg, you and I both know that you'll disagree with my numbers, just as I disagree with yours. In point of fact, both sets are pulled out of the air, and each client has to do the math for himself to see what makes sense for him. All I can say is that I have clients who are hosting their own email and groupware and are paying bupkis for it, having long since amortized the cost.

All that said, for some companies in some situations, some of your numbers make sense. For instance, the $10k cost of a server is typically used at the multinational where I frequently contract... but that's not an actual cost; it's an accounting ballpark used for budgeting and chargebacks. Then again, the meeting simply to determine the need for that hardware easily costs $500. However, these numbers are not appropriate to SMBs.

Likewise, there are really small businesses that do all of their accounting on a single desktop with Quicken or PeachTree, with no need for separate stations for order entry, shipping, or accounts payable. They may never have the need for a local server, and thus would really benefit from hosted email and groupware. Unfortunately, these are the businesses that are also prime candidates for Gmail and Google Docs. Unlike larger companies, their downtime is not measured in thousands or millions of dollars per minute.

Nevertheless, while recommending a hosted solution for them (whether Google's or something like yours where it makes sense), continued access to their data is still a risk, I'd still recommend to them that they set up the mail account with POP or IMAP so as to maintain a constant backup of the information on those servers. Ditto with the docs; so once again...
Don't leave your critical data in someone else's hands.

3 Comments:

Anonymous Greg Charland said...

Hi Dave,

Me again. I'm not going to mention cost this time. I will say that the viable business prospect for SMB tech consultants is changing. If I'm charging a high-end consulting rate I can't be a tech, I need to be a CIO.

Clients trust me to tell them the most cost-effective place to put their stuff. Just like they trust the bank to be a safe place to put their money, or a locksmith.

I'm finding that the numbers these days (in the SMB space where I spend most of my time) do not support on-premise mail servers unless there is extreme paranoia or advanced security requirements. I've been moving toward low-end file servers for authentication and line-of-business apps if needed, but e-mail/sharepoint/backups/etc are most often offsite.

Consider a typical "stacked" configuration I am very, very comfortable with: Inbound mail goes through Reflexion Networks' RADAR live archive + anti-Spam...then on to ExchangeMyMail hosted Exchange and Blackberry Enterprise Service.

If there is a problem I do about 18 seconds worth of troubleshooting then fire off a ticket to Reflexion or EMM and keep an eye on their resolution of the issue. I'm not the tech, I'm the CIO.

Let's walk through the "game" I play with my clients:
What's the worst that can happen?

-EMM goes under? All mail goes through Reflexion's live archive. It's a beautiful thing.

-Reflexion goes under? The mail is really on EMM's servers so I change my DNS MX entries and the mail flows again within a few minutes.

-EMM AND Reflexion both go under on the same day? We recover the .OST files, get new hosting set up, and pray we don't get fooled again.

-Major technical problem with Reflexion? Will never happen. If it did I would reset MX entries to EMM directly.

-Major technical problem with EMM? The users know they can log into Reflexion RADAR and work their inbox in real-time while EMM works on the problem.

-Major security problem with either? Not gonna happen. Such an event would put them out of business and they know it. Worst case: we notify clients and call in their E&O insurance.

Although failure of either of these companies is almost unthinkable these contingencies are part of my DR planning.

90% of on-site servers were set up years ago by a guy who didn't know what he was doing, did not configure automatic updates, no antivirus, and no backups. The business owner thinks that the on-site server gives some protection and security...which couldn't be further from the truth.

In the hosted world, these guys have really nice stuff that small businesses just can't afford. Like multi-homed net access, storage clusters, VMotion (or whatever it's called these days) and sysadmins that work on virtualized Windows 08/Exchange 07 ALL DAY EVERY DAY.

I choose my vendors carefully because I know we've got a lot of eggs in their baskets. An SBS or LFS server has a lot of eggs in its basket, too and will never have the same level of redundancy, reliability, and protection as a properly-configured hosting server...even when we follow best practices.

September 22, 2009 at 8:42 PM  
Blogger Dave Leigh said...

Greg, I'm glad that you go through the exercise with your customers. I have to do something very similar, as should ALL vendors, IMHO. As we've seen in this economy, no one is "too big to fail".

Disaster Recovery is an aspect I'm glad you mentioned, as it's far too often overlooked by SMBs, and it's one of the best arguments there is for paid hosted services. If your data are not on site, then they can't get washed away in a flood, burned down in a fire, or carried away by a burglar.

But while most people err in DR by not thinking about it at all, others err by putting those eggs in one off-site basket, not considering that they're not the only ones who can fail, or that "disasters" can also take economic or contractual form. Maintaining the .OST file mitigates that for some data, and I'm glad to see you're thinking in those terms. (For those that are reading, this refers to offline storage... basically an offline copy of your exchange data, similar to a Notes database replica).

Because Lotus Notes/Outlook provides so much more than Exchange in terms of flexible, customized solutions (and here I'm giving a blatant self-serving nod to my open source VIC CRM), I personally prefer it; and I still prefer to keep core data close, for reasons I've already stated.

I won't weasel-word it: I don't like cloud computing. I believe that critical data do not belong in others' hands. And I continue to encourage people to do the math for themselves.

That said, I'm not going to pretend that some people haven't actually done the math and found that, for them, solutions such as yours are a winner.

I sincerely hope that you sign up every client possible for whom this is true -- and not a single one for whom it isn't.

All the best to you!

September 23, 2009 at 11:31 AM  
Blogger Dave Leigh said...

More grist for the mill:
http://www.enterprisestorageforum.com/technology/features/article.php/3843151

October 10, 2009 at 6:24 PM  

Post a Comment

<< Home