Uptime Report - April (04/05)

[Trip]

Host-Tracker numbers for all included services as of April, 2005. HT got rid of response times, as the numbers, I'm guessing, were not credible. Here's the list, ranked on uptime:


KEY:
Company Name - URL Monitored
Uptime% - Access Failure Time - # Failures

GMail by Google - http://www.gmail.com
100.00% - 0 min(s) 0 sec(s) - 0 failure(s)

Yahoo! Inc. - http://mail.yahoo.com
100.00% - 0 min(s) 0 sec(s) - 0 failure(s)

MSN Hotmail - http://www.hotmail.com
100.00% - 0 min(s) 0 sec(s) - 0 failure(s)

Tuffmail - http://webmail.tuffmail.net
99.98% - 10 min(s) 9 sec(s) - 9 failure(s)

bluebottle solutions pty ltd. - http://www.bluebottle.com
99.95% - 20 min(s) 37 sec(s)- 24 failure(s)

Fadmail.com - http://www.fadmail.com
99.94% - 25 min(s) 59 sec(s) - 17 failure(s)

MailSnare.net - http://webmail.mailsnare.net
99.91% - 34 min(s) 47 sec(s)/27

FuseMail, LLC. - http://fuse1.fusemail.net
99.91% - 38 min(s) 49 sec(s) - 51 failure(s)

FastMail IP Partners - http://www.fastmail.fm/mail
99.87% - 52 min(s) 16 sec(s) - 29 failure(s)

PromptPost Project - http://mail3.promptpost.com/mail/src/login.php
99.65% - 2 hour(s) 26 min(s) - 81 failure(s)

BurntMail - http://www.burntmail.com
99.51% - 3 hour(s) 18 min(s) - 27 failure(s)

runbox AS - http://aibo.runbox.com
99.33% - 4 hour(s) 34 min(s) - 168 failure(s)

United Email Systems - http://www.unitedemailsystems.com
99.22% - 5 hour(s) 25 min(s) - 150 failure(s)

Cwazy.co.uk's Zapo.net - http://www.zapo.net
99.21% - 5 hour(s) 33 min(s) - 108 failure(s)

Virus Free Email - http://www.vfemail.net/horde/imp/login.php
96.07% - 1 day(s) 3 hour(s) - 148 failure(s)


***MONITORED SINCE 4/8/05***

Gawab.com - http://gawab.com/webfront/main.php
99.93% - 19 min(s) 31 sec(s) - 23 failure(s)

EuMX.net - http://www.eumx.net
99.91% - 25 min(s) 32 sec(s) - 21 failure(s)

BeanRack - http://www.beanrack.com
99.78% - 1 hour(s) 8 min(s) - 71 failure(s)


The "big three" were on top, and with such huge resources, they will probably stay there. April wasn't as hot for the the smaller-scale providers. Six of them were down more than an hour. A dismal month for Runbox (again), low numbers for BurntMail, and look out for VFEmail (more than ONE DAY of downtime, woah!). But Tuffmail finished best with only 10 minutes of access failure and bluebottle improved quite a bit.

Trip

[Edwin]

Sorry, I have to take issue with your "dismal" comment about Runbox, and indeed with the bottom-end of the table (VFE excepted). For a relatively low-priced consumer service, an uptime comfortably over 99% in what seemed to be a pretty bad month is very good.

In fact, I would go as far as to say that a 99% uptime should be a sufficient "quality bar" for any provider in today's market. If they can sustain 99% a month, month after month, that's very impressive.

Basically, the quality curve is asymptotic towards infinity - the more you want to increase the factor after the decimal place (".") the more the cost rises more and more steeply. This is well known in the hosting industry, for instance - tacking an extra 0.x% of reliability onto a service can double the cost of the service. Want more? Double it again, and again etc...

Also worth noting what the hosting industry tends to offer in terms of compensation - while the exact terms vary, it's usually a percentage of monthly fee refunded based on downtime if the uptime guarantee can't be met. For instance a hosting service with a 99% uptime guarantee but with real-life uptime of 95% would refund 4% of monthly fees for that month.

Google around for a bit if the above looks unlikely - you'll soon find out that it's the norm.

[ReuvenNY]

Trip, thanks again for your report! Can you please add to your list goowy.com and inbox.com? Those two seem to have arisen interest among the members here.

[rmns2bseen]

100% uptime for Gmail??

[gdg]

^I've never experienced a downtime with Gmail...but I don't try to access it 100% of the time.

[rmns2bseen]

I haven't had any problems either, maybe a minute or two here and there...my surprise comes from reading comments made by other users who apparently have problems logging in etc. But wow, this report after I had recommended to someone else in this forum that they should try VFEmail

[Trip]

Originally posted by Edwin
Sorry, I have to take issue with your "dismal" comment about Runbox, and indeed with the bottom-end of the table (VFE excepted). For a relatively low-priced consumer service, an uptime comfortably over 99% in what seemed to be a pretty bad month is very good.

I agree with you Edwin. That sounds a little harsh, but only because I actually use Runbox all the time, so when it's down I'm usually experiencing it. And Runbox had a few days where they were unreachable for 30-40 minutes at a time. So although statistically minimal, when your life pretty much revolves around email sent and received all day, you can definitely "feel" that much downtime, regardless of how it appears in numbers.

But yes, I understand your point "100%" (Pun intended).

Trip

[Chipper]

Hi Trip --

Thanks for the info -- very interesting and enlightening.

I don't want to go overboard with stats, but would it be possible to also list:
1) The longest outage
2) How many outages lasted longer than x minutes (don't know how many mins x should be — perhaps others can weigh in here)

My reasoning for tracking these two items is that "averages" can sometimes be misleading. 99% uptime sounds great (only 14 mins, 24 secs per day). However, that's just the average per day — if the entire outage happened all at once, that's about 7 hrs, 26 mins in a 31 day month — a little under 1/3 of an entire day! However, I don't want to create a lot of extra work for you.

-- Chipper

(and I'm 100 away from the "evil post count")

[registered_user]

First off, thanks Trip for posting this report. Secondly, does anybody know of any reporting service that could track how often email services wind up on blacklists? I'm very happy for 99% or better uptime, but it doesn't mean much when SpamCop causes my outgoing mail to bounce. I imagine that such a tracking service does not exist (and would be difficult to provide), but it would still be a cool idea.

[rmns2bseen]

100% uptime for Hotmail is really stunning. I take it the stats don't include inability to log on because the "servers are too busy"? That's an almost daily occurence with Hotmail.

[Trip]

Originally posted by Chipper
I don't want to go overboard with stats, but would it be possible to also list:
1) The longest outage

VFEmail: 2005-04-07 - 8 hour(s) 54 min(s)/30 (also 2 other outages of 8 hours for VFE...)

2) How many outages lasted longer than x minutes (don't know how many mins x should be — perhaps others can weigh in here)

How about we make it 10 minutes for now (thats about the average time of small upgrades or server reboots). I ordered them on severity of the outages:

GMail: 0
Yahoo: 0
Hotmail: 0
Tuffmail: 0
Bluebottle: 1 - 12m
Fadmail: 1 - 26m
MailSnare: 1 - 28m
FastMail: 1 - 43m
Fusemail: 2 - 28m, 10m
Prompt Post: 2 - 1Hr 2m, 1Hr 23m
Burntmail: 1 - 3Hr 18m
Runbox: 8 - 16m, 38m, 10m, 1Hr 12m, 10m, 33m, 1Hr 11m, 13m
Zapo: 4 - 1Hr 33m, 24m, 1Hr 27m, 2Hr 1m
VFEmail: 5 - 24m, 7Hr 49m, 2Hr 59m, 8Hr 54m, 7Hr 59m
United Email Systems: 7 - 23m, 1Hr 40m, 11m, 3Hr 8m, 45m, 36m, 19m

***Monitoring starting the April 8***
Gawab: 0
EuMX: 1 - 21m
Beanrack: 2 - 17m, 35m

My reasoning for tracking these two items is that "averages" can sometimes be misleading....

Absolutely. I think in the future I will just post the raw Host Tracker log in a stripped down form so that people can see when and for how long downtimes occurred.

Originally posted by rmns2bseen
100% uptime for Gmail??
.... 100% uptime for Hotmail is really stunning. I take it the stats don't include inability to log on because the "servers are too busy"? That's an almost daily occurence with Hotmail.

Not surprising for either service. Yahoo as well. Think of it in terms of resources. Those companies are so well endowed with probably hundreds of servers, which likely create higher redudancy than the smaller providers. Thus, as I said, you'd expect them to be "down" less, but as you said, the service itself could be lousy. A "server too busy" signal is not seen as "down" to Host-Tracker. but in that case the server load is so high that the system automatically stops allowing more sessions to be created by blocking more logins until the load decreases to a certain level. That's what you get for using Hotmail.

Originally posted by registered_user
...does anybody know of any reporting service that could track how often email services wind up on blacklists...I imagine that such a tracking service does not exist...

Actually, the following URL provides an aggregated list of ALL blacklists:
http://www.sdsc.edu/~jeff/spam/cbc.html
Also, when an email you send to someone is rejected, the the mail daemon should specify in the reply email (if it is bounced back to you) why or on what spam rule the message was rejected. If you can figure out the blacklist from this (ie. I've seen "RCVD_IN_SBL" enough to recognize that as a Spamhaus blacklist entry), you can go to the blacklist operator's website and check the suspect IP number to see the status. If it's listed, then your best course of action would be to contact the blacklist operators to try and get your IP removed.

Trip

[rmns2bseen]

thanks for the explanation, Trip and pssst never let it be said I use Hotmail on a regular basis

[xbot]

I see Beanrack had a good month . Cheers to hope for many more to come

[Shelded]

Originally posted by rmns2bseen
100% uptime for Hotmail is really stunning. I take it the stats don't include inability to log on because the "servers are too busy"? That's an almost daily occurence with Hotmail.

I don't have such dismal experience but the fact is, it happens. Too bad the month being measured is mere days old. Not a very statistical way of handling matters.

Originally posted by gdg
I've never experienced a downtime with Gmail...but I don't try to access it 100% of the time.

I very rarely use Gmail but happen to have caught it down more than perhaps they would wish. Maybe they have tried to increase performance since sending so many 50-invite packages out.

[Shelded]

Originally posted by Trip
Not surprising for either service. Yahoo as well. Think of it in terms of resources. Those companies are so well endowed with probably hundreds of servers, which likely create higher redudancy than the smaller providers. Thus, as I said, you'd expect them to be "down" less, but as you said, the service itself could be lousy. A "server too busy" signal is not seen as "down" to Host-Tracker. but in that case the server load is so high that the system automatically stops allowing more sessions to be created by blocking more logins until the load decreases to a certain level. That's what you get for using Hotmail.

Trip

So does this mean if there's a router down between you and Fastmail that Fastmail gets dinged for the outage?