Archive for the ‘servermojo’ Category

New Test Node – Kansas, USA

Sunday, June 7th, 2009

We have just added a new monitoring node to our test network, this time based in Kansas and sponsored and hosted by WiseDedi Internet Solutions (wisededi.com).

WiseDedi provide top quality managed and unmanaged dedicated and VPS servers at a price which is hard to beat.

This new node comes with 1G memory and enough CPU power to ensure the node runs smoothly all the time, as it is indeed doing now. It already performs countless tests each day and remains rock steady and highly responsive and still shows barely any load!

If you are looking for great hosting, check out WiseDedi now. We are very pleased to include them in our growing monitoring network and look forward to a long and mutually beneficial partnership.

We would like to take this opportunity to say that we are constantly looking for more testing nodes in new/different geographical locations. If any other hosts are interested in sponsoring test nodes, please contact our support team (see support for details).

Search Speed Shootout – Google vs Bing vs Yahoo vs Ask vs Twitter

Friday, June 5th, 2009

Well the web is buzzing with search news these days with the launch of Bing (which is actually not too shabby!) so naturally we decided to take a closer look to see which performs best in terms of speed. We setup monitors for the index pages and for actual searches (while checking for a valid search result) and the results are in some ways what you might expect but in others, quite surprising!

Without further ado, here are the results..

Search Speed

Err.. well a bit more ado.. We perform a search every 5 minutes and check that the keyword we searched for appears in the result page. Response times are average over the last day and the charts are ‘live’ so you can see what the current status is as you view this article. We can probably assume these search engines are using fancy caching of some kind but we don’t care.. now we are only interested in the result speed not how they do it. No doubt they all have their own ways to work their magic and return a good result in a timely manner.

Rank Search Engine Response Time
#1 Google 649ms
#2 Bing 665ms
#3 Yahoo 965ms
#4 Ask 1111ms
#5 Twitter 2115ms

No surprise that Google is the fastest search engine around (of those we tested) but look at Bing – just 16ms slower than Google! Also of note, Google “cheats” in a way since their results pages do not actually contain the results if you view the source so probably they use ajax or some other technical wizardry to load the results in the page after it starts loading. Bing does include the results in the page source so possibly this 16ms difference could easily be made up if Bing tried the tricks that Google uses.

Poor Twitter search is struggling at the back there taking more than three times as long as Google to return a result.. however to be fair, despite barrel loads of VC cash Twitter still doesn’t quite have the resources which are available to Google and Microsoft. Clearly though, they have work to do there!

Index Page Speed

Again we check every 5 minutes but this time only the index page. Results here were not what I expected at all! Yahoo in first place and the mighty Google trailing the field?! No doubt heads will roll at the Googleplex for that! ;)

Rank Search Engine Response Time
#1 Yahoo 748ms
#2 Ask 933ms
#3 Twitter 1017ms
#4 Bing 1062ms
#5 Google 1067ms

You can click the names of the search engines in the columns above to see the full live report for each which includes daily and monthly charts for speed and uptime/downtime and other exciting statistics.

So there you have it.. Google is first and last! Bing is snapping at the heels of Google with a very respectable result (so Steve Ballmer wont be launching any chairs just yet! ;) ) and Twitter needs to get their skates on.

Update: seems that Google does include results in the results page source sometimes, not sure if its random or based on some query string param or who knows what but interesting to note anyway. The brief downtime you see on the Google monitoring report on June 7th was just us testing the results page source – Google wasn’t really down so don’t panic! ;)

Alerts after..

Friday, May 29th, 2009

Today for “featurefriday” we added the ability to define how many tests must show a status change (up or down) before we send you an alert.

By default we send an alert immediately when we detect a change in server or monitor status but now you can set a number between 1 (the default) and 20. In other words up to 20 tests before we send you an alert if you choose. This can help minimize alerts triggered by transient network or server issues.

Enable this feature from your settings page – just look for the “Alert after..” setting and select one of the options from the dropdown list.

Timezones

Friday, May 22nd, 2009

Another new feature is now live on ServerMojo – timezones! This means you can now set whatever timezone you want your monitors to work with and your stats, charts and alerts will show the times based on your choice. This should make it easier to correlate servermojo monitoring events with your own internal systems or local timezone.

To use this feature just go to your settings page and select the timezone you want. This setting applies to all monitors in your account. The default timezone is still GMT/UTC and can be left as it is if you don’t want to localize it.

Custom Timeouts

Sunday, April 12th, 2009

On friday we released a new feature – custom timeouts! This is particularly useful on http type monitors and can be used to reduce ‘false alarms’ caused by a slow server response rather than a fully offline/unreachable server or site.

In many cases you may be monitoring a server which is very loaded or on a slow connection or just at certain times of the day slows down either due to visitor traffic patterns or running scheduled tasks. Since you are already aware of this status you don’t want to be getting alerts all the time saying your server is down.. so now you can simply define a longer timeout (the system wide default is 5 seconds) and this should eliminate such alerts in all but the most extreme cases.

To set your timeout just visit your settings page and enter any number between 1 and 20 seconds. This setting applies to all monitors in your account. To decide what timeout is best for your case just look at the response time charts for your monitors over the last 30 days and then choose a time slightly above the typical response times you see there.

Let us know if you have any questions or suggestions.

Shooting Twitter – perils of real time search..

Friday, April 3rd, 2009

Sad news is just coming in now of a shooting in Binghamton in the US.. several dead and injured.

Of course it immediately became a very hot topic on Twitter.

You can see the effect it appears to have had on twitter systems with a sudden big increase in page load times to the point of being basically down now to all but the most patient..

Twitter Response Time

Twitter Downtime

Hopefully they’ll throw in some more computing power or otherwise solve the load issues soon. You can follow twitter server up/downtime here..

http://www.servermojo.com/user/10000/monitor/10682/

Obama is DOWN!

Tuesday, March 24th, 2009

I just got an alert via twitter, sent from ServerMojo telling me the presidents web site is missing!

Here’s todays chart showing the angry red downtime clearly..

Obama.com Downtime

The downtime lasted from 17:35 to 18:25 (GMT) so for 50 minutes, obama.com was MIA.

As it is now back online, the world can breathe a sigh of relief but had those folks at the whitehouse been using ServerMojo, perhaps they’d have fixed it sooner.. ;)

You can follow the status of obama.com here..

http://www.servermojo.com/user/10001/monitor/10230/

Twitter aflutter..

Friday, March 20th, 2009

Seems like Twitter is having a bad day with multiple slowdowns and timeouts over the last 24 hours as can be seen on the following charts.. (all times are GMT)

Twitter Uptime (last 24 hours):

Twitter Response Time (last 24 hours):

As you can see the down periods correspond with spikes in response times meaning the page fetch timed out when we checked it – we have a default 5 second timeout for http tests and when a page (html only) takes over 5 secs to download its not generally a good thing. As a result and at the time of writing this blog post their uptime for the last hour is down to 89.86% and 97.19% for the day which for a system as popular as Twitter is sure to annoy many. In fact a quick search shows many complaining tweets already and it’s only a matter of time before they get raked over the coals again by Techcrunch! ;)

Compare that with the last month and you will see that its usually much better even though they do still have the occasional dip in performance..

Twitter Uptime (last month):

Twitter Response Time (last month):

In fact twitter uptime over the last month has been a fairly respectable 99.76% and for the year so far it is 99.85% so not too shabby overall and the infamous fail whale has not had as many opportunities to show itself recently as it did last year!

Just a reminder – not only do we monitor sites but we also provide alerts direct to your twitter account! You can follow us here: http://twitter.com/servermojo to keep up with the latest ServerMojo happenings in 140 characters.

** UPDATE: its been a couple of hours since I first posted this and there is clearly an ongoing performance issue at twitter today – the last few hours have been significantly slower than usual and in fact the average response time for today has reached 1211ms. Compare that to the time for the year so far which is only 633ms, for the last month 850ms and last week 898ms, so its quite a change and in fact appears to be trending towards increased slowness in general. So what is going on at Twitter HQ today? Is it the increased visibility of the search link resulting in their servers getting hammered? Or is there some event going on now causing a big increase in tweets?

New Test Node – Madrid/Spain

Thursday, March 19th, 2009

We have just added a new test node to our test array, this one located in Madrid, Spain.

It is also our first sponsored node, sponsored and hosted by cyberneticos.com, which offers both dedicated and VPS hosting. 
Be sure to check out cyberneticos.com for more information about their hosting plans and hosting facility in Spain.

Already fully operational, the testing node at cyberneticos.com, which is a VPS node with 512MB-1GB memory, and a decent CPU share is easily operating at an extremely low load and CPU usage, which is a very good indication of the server’s status. Naturally we keep monitoring our own testing nodes. :)

cyberneticos.com is also a ServerMojo user and came up with some nice feedback about sub-accounts and other UI aspects that were implemented. Thanks for the feedback and the server, Anthony.

We would like to take this opportunity to say that we are constantly looking for more testing nodes, in new/different geographical locations. If any other hosts are interested in sponsoring test nodes, please contact our support team (see support for details).

Public Stats Pages

Monday, March 9th, 2009

Existing users may have noticed that on the settings page there is an option for public or private stats. What you may not have known (since we didn’t post it anywhere) was what your public stats urls actually are – until now.

Now you will see the public stats url displayed on all the relevant pages so you can share this url or post it anywhere to show other people your server uptime stats. Thanks to CogDogBlog for reminding us! :)

If you have any questions or suggestions, give us a shout.