Software

Email servers are a commodity. Email contents are not.


Note: this was originally posted as a comment (http://www.zdnet.com/5208-10532-0.html?forumID=1&threadID=17796&messageID=350047&start=-1) to David Berlind's article Yes. You should outsource your e-mail

Mr. Berlind is absolutely correct is quite a large number of his statements, and he does indeed provide a compelling arguement for outsourcing email. But he has made some mistakes, which is where we differ on this topic.

The first number one problem here, is that Mr. Berlind's original blog post is titled "Google to provide email hosting?" and is 100% about outsourcing your email to Google. I put forth the question "Now, let's look at the premise: assuming I would outsource my email, why would I outsource it to Google, of all companies?"

Mr. Berlind has not even addressed this at all. Not in the slightest. He provides a good (but flawed) arguement in favor of outsourcing. He does not even touch the idea that Google should be the one to do it. Mr. Ou adds a quick little list of why, even if outsourcing email is the right choice for your company, Google is not the one to do it (http://www.zdnet.com/5208-10532-0.html?forumID=1&threadID=17781&messageID=349545&start=-1). Admittedly, he did not go into nearly as much length or detail as I would have, but his comments really don't need much explaining (except for the user interface bit; GMail is pretty decent as far as web mail goes, and is garbage compared to a desktop app, is a good way of putting it).

Now, onto the topic of the current blog post by Mr. Berlind: "Yes. You should outsource your e-mail".

I say, "No. You should NOT outsource your e-mail".

"So, one question I have for Mr. James is, of all the stuff being outsourced today, what of it isn't mission critical?"

That all depends on the business, but I have not encountered a business that did not consider email to be mission critical since about 1998, if not earlier than that. Furthermore, the fact that companies *are* outsourcing portions of their business process does not mean that they *should*. I can think of a number of anologies to this, but this principle is best summed up by David Hume: "One cannot derive an 'ought' from an 'is'". If everyone in New York jumped off the Empire State Building, that certainly does mean that I should as well.

"But aside from the handful of hand-built customized competitive advantage-driving systems that integrate messaging and email into their functionality, are any of us really that deluded to believe that insourcing something as basic as email can make us more competitive than the next company (setting aside those companies with real security concerns that can prove their insourced system is more secure than the outsourced one)."

This is a correct statement on the technological level, but an incorrect statement on the business level. On the business level, it is not the email system itself that matter; if carrier pigeons were ferrying letters printed up on Guttenberg presses at the speed of light, businesses would use it. What matter on the business level is the contents of the email, that is the mission critical part of email. Email is the lifeblood of companies, having replaced to a large extent phones, couriers, postal systems, fax machines, and so forth. What is contained in an email is often of an incredibly sensitive or important nature. Furthermore, archived emails are a frequently a knowledge repository. There is a reason why my personal email archives reach back to 2000 (and would go back to 1996, if I did not have a lapse in judgement in 2000). My major disappointment with email is that the tools are still rather primitive for mining that data.

Any outsourcing situation has to acheive at least one of two goals: better value, or less cost. There are two ways to handle outsourced email: one is the way most small business do it, to have an external host, and they pick up the mail via POP3 and store it internally. The other way is to have an external IMAP or Exchange server offsite, and leave the data there. In the first situation, all you have outsourced are two TCP/IP transactions, one for SMTP and one for POP3. There is no added value here. And there is no reduced cost. If this is all your needs are, get the cheapest server you can find, load a Linux or BSD on it, and load qmail. For $500 in hardware, and the recurring fees for DNS and domain name registration, you are providing the same level of service to your company that the external provider is. Heck, your existing Windows server (nearly every business larger than 10 employees has one now) comes with an SMTP and POP3 server on it. So outsourcing this level of email service adds zero value and costs a lot more. Use those, if you don't feel like getting a second server. The second option also adds no value (again, you can put Exchange onto a server yourself) and costs more. Look at the numbers you quote from Centerbeam: $45 per user per month. In a company of 10 people, that is $5,400 per year. That is more expensive than a server with Windows 2003 and Exchange, plus a lot of data AND a backup solution! Go the open source route (didn't you guys just blog about Scalix a day or two ago?) and you have enough money left over to buy every employee an XBox 360 for a bonus. Gee, that doesn't seem like such a value at all, does it?

"What we offer to do is the hard work for people that they can't afford to do themselves."

I know this is you quoting someone else, so I am now arguing with them, not you. As I show above, anyone who isn't working on the US Government's budget can see the math problems here. How much does it cost to run your own SMTP/POP3 server? It takes what, a few hours to properly setup and establish a server using either *Nix with qmail or Scalix, or Windows with the built-in servers or Exchange? The Windows route is especially useful, because all of your account management is being handled via Active Directory, so that is one less system to learn. Is an outsourced server, regardless of what it is (POP3 or IMAP/Exchange) going to integrate with your in-house identity management system? I think not. What, you're going to set up a PPTP connection to their system, do a trust delegation to their AD system and yours, just so you don't have to manage separate usernames and passwords? Or would you prefer some awful webadmin system to go in and change stuff?

"That covers desktop management (anti virus, backup and restore everyday, 24/7 800# dial up helpdesk, server management, email management, VPN services, etc.)."

The in-house solution, except for 24x7 support, is still cheaper. Sorry.

"All a banker wants is more bankers and salespeople on staff. They don't want a Microsoft Certified Exchange Engineer on staff who is only available for one shift a day.   Even if you do run an Exchange Server with three shifts of engineers 7 days a week, they'll be advising you on best practices such as backup and restore. They'll say you need a Storage Area Network (SAN) and need to send tapes to Iron Mountain everyday."

This guy makes me forget just about every bad example I have ever given in a ZDNet TalkBack. Bankers have all of these things anyways. Bankers run a 24x7 database of millions if not billions of database entries where even a moment's worth of downtime can cost millions of dollars. This organization is going to be unable to support an additional few servers for email? But let's pretend he didn't say "banker". Let's pretend he said "small business owner". If having these hordes of MSCEs on staff is a problem for him, he may want to check out *Nix+qmail. I personally cannot vouch for Scalix (never used it, relatively new), but *Nix+qmail is a time tested, battle hardened system. It requires zero maintenance. None. Heck, Exchange, when properly configured, doesn't need any maintenance anyways. And at the end of the day, what good is his elite commando team of MSCE's going to do for a business owner if that business does not have someone on their end who can actually understand what they are saying and how to work with them? The only time his MSCE army is a decided advatnage is when their server starts behaving erratically and the problem is definitely on their end. If their software does something like that, where you need an MSCE to troubleshoot something that was working fine, then maybe that software isn't very good.

"[T]he point is that a leveraged model (where an outsourcing outfit spreads the infrastucture costs across more users than you can) is not only going to save you a lot of money, but headaces (sic) too."

Economics of scale is an idea I can buy into. But if they are leveraging economics of scale so well, why do they need to charge $45/user/month? Earthlink charges me $6/month for a POP3 only account. Is Centerbeam's economics of scale really so bad that they need to charge nearly 8 times as much for Exchange services? Maybe I need to reconsider those Exchange servers at my company, and put my new BSD server there to task with the email duties, the idea that Exchange is 8 times more expensive (with economics of scale, so it must be a few dozen times more expensive for our 5 person company!) than basic POP3 is total hogwash. Economics of scale distributes the cost of a Windows & Exchange license to something like 25 cents per user. So they're just ripping you off. Sorry, I don't like to be rpped off, and neither does my boss.

And what headaches are they really solving for me? Managing and maintaining an email server? It seems to me like they are giving me new problems, not taking away any existing ones. Let's make a headache list:

In-house:
- Hardware failure
- Network failure (immediate Internet connection and LAN only)
- User maintenance
- Initial installation and configuration
- Backup/restore
- Patching
- Security (95% of this is a subset of install/config)

Outsourced:
-/+ Hardware failure (should be be a problem if they are doing their job right)
- Network failure (their network AND immediate local Internet connection and LAN only, we've doubled our headaches)
- User maintenance (compunded by it not being part of my local authentication scheme)

So really, all I am giving up is responsibility for the hardware, backups, and maintenance. As I have already stated, email servers require nearly zero maintenance. I am patching my internal systems anyways, adding one more item to the list. Backups, again, I should be doing this anyways, what's one more server to have dump to tape/SAN/NAS? And for this I would be paying $45/user/month, which is much more expensive than a single Windows server in a 10 person environment? And if I have a large company, I can apply economics of scale to myself. An Exchange server can handle 1,000 users without a problem. Can my IT budget handle $45,000 per month for email alone? That's the cost of adding 6 MSCEs to my staff on a full time basis! And then I could have two of them monitoring my servers 24x7. Hmm, maybe I had better stop discussing Centerbeam's business model before his investors pull out now. Or better yet, maybe I should call those investors and ask them if they'd be interested in this bridge I have to sell, it connect Brooklyn to Manhattan...

"Raise your hand if you've used GMail, Yahoo Mail, AOL's mail or HotMail because you needed to send mail but couldn't get access to your corporate email system (for whatever reasons)."

I'll agree with you on this one. It's happened to all of us. On the other hand, this is an apples-to-oranges comparison. If I had the money to pay for Centerbeam's services, I would have the money for redundant email servers, in which case the only thing that would take me down would be a disaster (natural disaster, virus/worm, fire, etc.) or a complete network outtage, in which case my users would not be able to reach GMail, HotMail, etc. or Centerbeam's servers. So once again, this flounders on the cost issue. And also remember, I'm comparing Centerbeam's Exchange servers to in-house Exchange servers. If you compare a plain old outsource POP3 to an in-house *Nix server, the numbers are even more in favor of the in-house solution.

"Email systems, as it turns out, aren't that easy to run 24/7."

I have been doing it for years. The only thing that ever goes wrong are things that take down an entire server, or the whole network. Again, the price of in-house vs. outsourced makes this point hard to argue.

"Lastly, for a commodity system like email, what leverage do you have over your certified email engineer to keep the email systems up and running 24/7?  His or her job? Oh, that's what you want.  You'd rather spend time hiring and firing email engineers than making money for your company? Service Level Agreements (SLAs) are a lot easier to negotiate and enforce with service providers than they are during an employee's annual review."

Ah yes, my most favorite topic in the world! I guess I *do* have to rehash this topic. OK, time to brew a fresh pot of coffee!

First, some links to the extensive library of my thoughts on this subject:

* http://www.zdnet.com/5208-10532-0.html?forumID=1&threadID=16070&messageID=318442&start=-1

^^^^^^^^^ Number one most important post about the subject

* http://www.zdnet.com/5208-11406-0.html?forumID=1&threadID=16244&messageID=321399&start=-1

* http://www.zdnet.com/5208-11406-0.html?forumID=1&threadID=16278&messageID=322734

* http://www.zdnet.com/5208-11406-0.html?forumID=1&threadID=16324&messageID=324436&start=-1

* http://www.zdnet.com/5208-10532-0.html?forumID=1&threadID=16070&messageID=318442&start=-1

Wow! There's a lot of real-world, real-life, in-the-trenches experience in those links!

Now, to be fair, I don't always think that outsourcing is always bad, indeed, I have presented a compelling business case for it under certain circumstances: http://www.techrepublic.com/5254-6257-0.html?forumID=99&threadID=184332&messageID=1921068&id=2926438

My direct response to Mr. Berlind's statement. Have you ever worked someplace and had a hard time with the customer, and the boss pulled you aside and said, "look, I know you're right and the customer is wrong, but we have to swallow our pride and give them what they want"? I have. That's the way companies work, as long as it is profitable for them. As soon as giving the customer want they want is no longer possible, they say "no". If a company cannot deliver on SLA (no measurability, no proof of failure, little enforceability, blah blah blah TPVs stink blah blah blah, just some self deprecation there at this late hour), you are tied to them for a contract. And what are you going to do? Spank the CEO for being naughty? It isn't like the underpaid, underexperience, fresh-from-working-at-McDonalds-but-know-how-to-setup-a-CounterStrike-server kids who fill Third Party Vendors are going to be held responsible if a customer is lost. I am a big fan of "The Buck Stops Here". TPVs always manage to find a reason why it isn't their fault, SLA wasn't truly violated, etc. For a TPV, "The Buck Stops Here" really means "Your Money Ends Up In Our Bank Account".

Employee annual reviews are a lot easier to manage than SLAs. I can directly measure and manage my employee's success. SLAs are notoriously difficult to manage. I have seen cases where a customer spent nearly as much time and money simply managing SLA than they did to manage the service themselves. That is rediculous. If you think SLAs are easy to manage and enforce, try an experiment: call your cable company to make a service call. You will get a 4 hour time frame where you must be home (heaven forbid if you're in the bathroom when they come by, "The Cable Man Knocks Once" would be a good film), and chances are they will be late anyways. If you're lucky, they'll give you some excuse about it. I remember working for a TPV and being instructed by managers to "invent" weather conditions that caused SLA misses, since poor weather was an SLA escape clause. How much money will you spend just hiring lawyers to 1) write the SLA 2) help you get out of the contract when SLA keeps being broken 3) sue to recoup the costs to your business when SLA is blown? If I have a bad employee who makes a serious goof, I can take them to task, or even fire them if need be and replace them with a more competant person. If SLA is blown, there is no recourse.

Finally, there is the issue of commoditization itself. Declining levels of quality are the largest result of commoditization, outside of pricing. Look at cars. The only reason why cars improved one bit after 1972, is being foreign competition started selling better cars at a better price in the 80's. Before that, American cars had become commoditized to the point where they were all equally junk. Now, American cars are often significantly better than their foreign counterparts, because they were forced out of commodity status. Consumer electronics is another example. Even though cell phones are cheaper now than five years ago, I spend more on them because their quality stinks. My year-old cell phone has worse battery life now than my friend's 5 year old analog phone. The worst thing that can happen, outside of a destructive monopoly, is commoditization. It freezes the desire to improve quality and replaces it with ruthless cost cutting to match the price cutting. When you cannot compete on features or quality because everyone is the same, then no one cares about it. Again, cell phones. At this point, consumers expect poor service, because that is the price we paid to save money. There are no "premium" or "luxury" carriers out there (well, Verizon is a bit pricey, and they do seem to have slightly better coverage, from my experience), but in general, cell phones stink. Why? Because with today's price slashing, no one can afford to innovate.

J.Ja

About

Justin James is the Lead Architect for Conigent.

0 comments

Editor's Picks