dcsimg

Rethinking Gmail: Reliability Matters

Gmail has changed the way we think about email, mostly for the better. Lately, though, Gmail has become notorious for its problems rather than speedy search and its "never delete mail again," storage allotments. It might be time for a tough love approach.

For better or worse, Gmail is one of the poster children for Web-based applications. Millions of people use Gmail to manage their email and depend on Google to provide reliable access to their mail. The problem is, Google has had a few high-profile outages that should be causing some concern about hosting anything mission critical with the Internet giant.

Another Day, Another Outage

This week, Gmail suffered yet another outage affecting an unknown (outside of Google, at least) number of users. This is hot on the heels of a massive outage that affected pretty much all Gmail users — including those of us who pay Google for the “Apps for your Domain” service.

Having worked in a data center, I have some sympathy for a company when there’s some downtime. Murphy’s Law can strike with a vengance, and I’m convinced that there is no such thing as enough redundancy to fully prevent service outages completely. What’s important is that those outages are minimal, managed well, and learned from.

Google may well be learning from its outages, and it seems to be managing them OK, but we’ve surpassed “minimal” at this point. The previous outage affected the entire service. This time, at least in my case, I was able to reach mail but unable to use my contacts and Gtalk became extremely unreliable.

Mail was available via IMAP and POP3, apparently, but isn’t the whole point being able to reach the service via the Web and have access to Gmail’s search capabilities? I’d be surprised if the majority of users have a client configured to access Gmail through IMAP or POP3.

Tradeoffs

When Google launched its “Apps for your Domain” service, with service level guarantees for $50 per year, per user, I was quick to move my mail from Tuffmail to Google Apps. The Gmail interface is vastly superior to other Webmail interfaces, and search is a key feature in the way I use email.

But it’s only superior if I can actually reach the service. For some users, it may be acceptable for email to be unavailable for an hour, or even a day. But for many use cases — like mine — email is mission critical. Having no way to access email for an hour can be a major problem in some cases.

You Get What You Pay For

The majority of Gmail users aren’t paying a cent for Gmail, so it’s easy to argue that they shouldn’t be looking a gift horse in the mouth. Or is it? It bears some careful thinking.

Google’s business model is dependent on happy users who continue to view, and occasionally click on, ads. Saying that the users aren’t paying is to discount the value that they bring to Google. If Google doesn’t keep those users happy, it risks losing them to a competing service and having less value for its advertisers.

To put it another way, Gmail’s users aren’t the customers, they’re the product and Gmail is simply a tool to harvest their attention on behalf of its advertisers. This is a symbiotic relationship so long as Google keeps up its end of the bargain by providing reliable services. So the argument that Gmail is “free” and that users shouldn’t complain about outages rings a bit hollow for me. Google needs its users as much, if not more, than the users need Google.

Migration

Migrating email providers is a headache. Doubly so when you maintain your own domain. Even so, Gmail’s recent problems have convinced me that it’s probably time to rethink using Google Apps as my primary mail service.

I’ve no doubt that one of these days, Google will be able to scale its service and provide more robust uptimes to its customers. In general, I think that Web based services will continue to approach “fat” client reliability. But we haven’t reached that state just yet. in the meantime, if email is something you depend on, Gmail may not yet be ready.

Comments on "Rethinking Gmail: Reliability Matters"

stevemadere

You must be some amazing system administrator to outdo
gmail for reliability.

Most companies do not have such amazing system administrators.

Most IT departments have no chance of matching the
robustness of Google.

Articles like this one sadly give them invalid examples to argue
otherwise. At many companies, (in particular those running exchange)
email outages are a monthly or even weekly occurrence and
they NEVER have decent search capabilities.

Reply
corktowner

Outage? Really? I had no such problems, but then again my life does not revolve around email and search capabilities. As for apps, I rely on OpenOffice so no problems there either. What are you raving about?

Reply
dperrin@unc.edu

I\’ve seen Novell\’s main site off-line for longer than that. Really, if you can outperform Google\’s uptime more power to ya. It\’s just that I don\’t believe you!

Reply
adriawave

And you have like some better service to suggest?

Reply
bofh999

This time i totally agree with the author…

First its easy to have a more robost system. make2 small cluster on different locations and sync every data …
google cant do that because of the massive data they would have to sync.

second big services like that have, naturally, more issues than small company based.

I dont say that google is bad. seeing the amount of users they have a hell of a job but its true that their users are not the customers, not even threaden like one. and dont even think about a real support.

theres also another issue you should really think about (specially all users who like to share their spreadsheets and so on)

big datacenters has many emplaoyments and youll have no idea who can access your data.

second a breach in their system will affect another unknown number of users and their data.

ok if you do only your household money management on a google spreadsheet noone would care. but mission critical information in shared open hostings?

sorry i dont like the idea.

btw you dont know if not any goverment whatever .org has (because of another more or less suspiciouse antiterror law) their agreements with those big data hosters like google or other..

you maybe think i should find some prof. help but remember the encryption issues in the late 90s and rethink.

i dont say it is or will be. but again. in such evoirements like google, windows live,.. you cant know. you cant controll, youve no clue.
not about the security of your data, not about the uptime of their service,.. about nothing…

for private use it might be ok but the author is talking about mission critical and here .. agian … i dont like the idea

Reply
jsilve1

Wtf? I have to agree with the majority of the commenters here: Gmail has bad uptime? Simply put, Gmail uptime is AMAZING. An outage, even that 90 minute one a coupla weeks ago, added in to their overall uptime is un-effing-believable. Seriously, how is any single sysadmin gonna be able to outperform that?

I don\’t know what the author is complaining about.

Actually, more generally, and speaking as a sysadmin, I can\’t get over the level of expectation that people have for Internet-based services in general. 1 minute of down time is unacceptable in the public eye these days. Computers are *machines*, people! They break on occasion. Actually, pretty frequently. They are complex to fix and set up! I would love to see a general awareness push towards getting the public to understand that the internet/WWW was created by humans, runs on machines that can and do break, and by several laws of nature, must be expected to stop working occasionally.

But now I\’m just venting. Thanks!

Reply
ctryon

One fallacy people often fall into is illustrated by the concept of the \”100 year flood.\” The idea is that you only see this level of water damage on average;, once every 100 years. However, even with that average, it\’s still possible to see two \”100 year floods\” back to back in two successive years.

Google has had some pretty fantastic up time over the past couple of years. Somehow, they\’ve had two outages in quick succession. That\’s still bad for Google, but hardly what I would call earth-shattering. I don\’t know what their service level has been over the past several years, but I\’d be willing to bet it\’s still a LOT better than most internal Exchange based email services.

(Another analogy might be the difference between traveling across country by car or by airplane. Statistically, air travel is much safer than traveling by car, but when a plane crashes, EVERYONE hears about it. Do people feel safer because they have the illusion that they are in control of their own destiny when they drive a car, or when they administer their own mail servers?)

Reply
aotto

The comments on this post are right-on. You must have realistic expectations. More on my blog:

http://adrianotto.com/2009/09/scale-complexity-reliability-support/

Reply
jfblaze

First, I\’m not an AMAZING system admin. We haven\’t had an email outage in two years and we serve over 10,000 users corporate wide. That is not to say there aren\’t problems.

Second, my private email hasn\’t had an outage in 3 years.

I expect Google to provide at least 99.99 uptime. That\’s only minutes of downtime each year. 2-9\’s of downtime really isn\’t acceptable.

I don\’t think it is unreasonable to expect that google can keep email online.
I don\’t agree with the author jumping ship because of this 90 minute outage.

The question is really what uptime is acceptable?

Reply
dwolsten

I have to echo adriawave\’s post: you have some better alternative to suggest? I\’d like to hear it.

Like many people, one big requirement is that my email be web-based. I can\’t access my home computer from work, and I want to be able to read my email from any computer, and not just wait until I\’m home. That would be like going back to a landline.

So, this confines me to webmail services. And this brings up the second problem: most webmail interfaces SUCK. They\’re horrendously slow, they\’re hard to use, they\’re slow, they make it hard to find specific emails, they\’re slow, you get the point? Gmail, while not as fast as a local email client (though it\’s probably faster when searching), is very fast compared to everything else out there, and the interface is excellent. It\’s so good that it totally changes the way I use email.

So if it goes offline for 45 minutes once per year, I can put up with that in exchange for the massive increase in productivity, and that MASSIVE savings in time (due to the lack of slowness) compared with competing webmail services.

Reply
jvdongen

Sounds all too familiar; here are my experiences: http://www.tholis.com/news/bye-bye-google-apps/

Reply
poncha

@stevemadere
Noone needs to or intends to beat google in terms of robustness. Or reliability for that matter. But noone is going to try and beat them in customers amount. You don\’t need to be a superhero to beat them hosting one (your company) corporate domain, however, as running one mail server for one single domain is an easy task. So in terms of providing a reliable mail service, you can beat them for your one single domain – easily. Oh, and do you think Google workers are superbeings? They are not ordinary mortal men? Think again.

@dwolsten
According to google apps status dashboard, only the LAST outage lasted an hour and a half until full service restoration. So we\’re far beyond the 45 minutes once per year.

Now, I must admit that they provide service beyond ordinary mail, but this doesnt mean that the service *quality* should suffer…

Reply
greggster

I have run my own email for about 2 years now and had downtime – some my fault, and it affected at most a dozen folks when down. Gmail going down is like a power outage in the bay area – a lot are affected at once, and to add to the angst, there\’s is not a local admin to talk to about it. Their uptime is stellar – previous comments are right, but its the width of impact that makes it sting so bad.

Reply
mike_r

Gee, Talking about uptime, and yet the Opensuse mail servers were down for an entire weekend due to a power outage. No backup generators or offsite backup?

It sort of makes Google look pretty good.

Reply
jzb

To those who think the expectations are unrealistic: Google is trying to convince enterprises that they can outsource their email, etc., to Google. This is far different than a community project that provides email aliases for members. I\’ve used other services that have had little to no downtime — and *no* unplanned downtime I\’m aware of — over the course of a few years.

As I said at the beginning – I\’m somewhat sympathetic to hosting problems, but the amount of downtime Google has had lately is unacceptable if you\’re dependent on mail. If you\’re a casual user of email and don\’t need consistent access, then it\’s not a problem. As a use of Google Apps for your Domain, Google has already this year exceeded its SLA for downtime for a service I pay for. And the outages were unplanned, which makes it worse. (I also pay for rsync.net and VPSLink services — neither has had an unplanned outage I\’m aware of in the last two years, but both services have had announced downtimes which I was able to plan around.)

Reply
redmumba

I\’m just curious; if Yahoo! Mail or Hotmail were to go down for an hour, would there be millions of Tweets and hundreds of news articles proclaiming it? Because it sounds like GMail is getting so much attention because its, well… GMail. Up until this recent 90 minute outage, such wide-spread disruptions were basically unheard of.

Like everybody else has said… you can\’t criticize GMail\’s uptime if you can\’t say something\’s better. In other words, 90 minute uptime or not–GMail is still highly reliable, and I would definitely challenge anyone to indicate something better (in terms of uptime).

P.S.: One thing to note is, most of the time, these problems only affect select numbers of people, not everybody. And usually only the web interface is disrupted; often times, IMAP stays consistent throughout these outages.

Reply
luuk.paulussen

What I find ridiculous is that you think you can\’t live without email for an hour. It\’s not like it\’s keeping the blood moving through your veins.

Reply
lostinmissouri

Guys, do the math four 9s of uptime is 52.5 minutes a year, three 9s is 8.76 hours. If you were in a different time zone say Europe you would have never known about it. And any exchange admin that can service millions of customers at this level of service please step forward or go do your updates and restart. Linux/Unix can get very close with perseverance and a good hardware budget and just about any of the enterprise software packages. And by the way, who made the rule that email is garanteed delivery, I must have missed that one.

Reply
zeroshade

The question here is how many years has gmail been up without any downtime at all? The 9\’s of uptime is based on an AVERAGE uptime. So (to use the post above me) four 9\’s of uptime is 52.5 minutes a year. Say you\’re down for an and a half one year, but not down at all for the year after. That\’s still four 9\’s of uptime because on AVERAGE you\’re down for less than 52.5 minutes a year.

How many years has gmail been up without any down time? seriously people the outage has been blown WAY out of proportion just because it affected so many people. If gmail can be up for 4 or 5 (? not sure) years without any downtime, then has about 2 hours of downtime in one year, thats pretty damn reliable if you ask me.

Reply

This is really fascinating, You are an overly skilled blogger. I have joined your feed and look ahead to in quest of extra of your wonderful post. Additionally, I have shared your website in my social networks

Reply

I every time emailed this weblog post page to all my friends, because if
like to read it then my contacts will too.

Reply

I tend not to leave a lot of remarks, however after browsing a bunch of comments here
Rethinking Gmail: Reliability Matters | Linux Magazine. I actually do have 2 questions for you if
it’s okay. Could it be just me or does it seem like
some of these remarks look as if they are written by brain dead visitors?
:-P And, if you are writing at other social sites, I’d like to follow you.

Could you list of every one of your social networking pages like your Facebook page, twitter
feed, or linkedin profile?

Reply

Very good post. Great.

Reply

Have you ever thought about adding a little bit more than just your articles? I mean, what you say is fundamental and everything. But imagine if you added some great visuals or videos to give your posts more, “pop”! Your content is excellent but with pics and video clips, this website could undeniably be one of the very best in its field. Fantastic blog!

Reply

Oh my goodness! an amazing article dude. Thanks Nevertheless I am experiencing concern with ur rss . Don’t know why Unable to subscribe to it. Is there anyone getting an identical rss problem? Anyone who knows kindly respond. Thnkx

Reply

Throughout the great scheme of things you receive an A with regard to effort and hard work. Exactly where you actually lost me was first on the particulars. As people say, details make or break the argument.. And that could not be more true in this article. Having said that, let me tell you what did give good results. The article (parts of it) is actually very engaging which is most likely the reason why I am taking an effort to opine. I do not make it a regular habit of doing that. Second, even though I can see a jumps in reasoning you make, I am definitely not confident of how you seem to unite your details which make your final result. For now I will yield to your position however hope in the near future you actually link the facts much better.

Reply

I always was concerned in this subject and stock still am, regards for posting.

Reply

I do like the manner in which you have presented this specific difficulty and it does indeed provide me personally a lot of fodder for consideration. Nonetheless, from just what I have personally seen, I simply wish when the reviews pile on that folks remain on point and in no way start upon a soap box regarding the news du jour. Yet, thank you for this superb piece and though I do not concur with the idea in totality, I respect your viewpoint.

Reply

Hello my loved one! I wish to say that this article is amazing, great written and come with approximately all vital infos. I would like to see more posts like this .

Reply

Real instructive and great bodily structure of subject material, now that’s user friendly (:.

Reply

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>