INTERDUBS status

For the evening of January 20th we had scheduled to do some structural maintenance.

During the preparation for this maintenance a hardware failure tripped a consistency check that automatically took parts of the system offline to secure data integrity.

We decide to perform the maintenance scheduled for later during the day in parallel to the checks that the hardware failure had triggered.

All systems came back clean and normal. Parts of the system where inaccessible for some clients between 5:30pm and 6:50pm PDT. If you, or your clients were impacted then please come forward so that we can credit you for the entire month of January.

We offer 100% uptime guarantee to our clients. When we do not deliver on this promise then your entire month is free.

On December 12 we crossed the 5 million file mark.

No interruptions or downtime in December in any of our 3 data centers.

Since all browsers have reached a decent level of support for modern features (aka HTML5) we started
to implement those in more crucial parts of the admin interface.
(Client facing pages still support pretty much every browser released in this century)

Yeah for progress!!

Yeah for 2012.

Thank you to all our clients and collaborators to make it an awesome year.
It was worth the 584 million miles the earth had to travel.

November saw new features.
November saw new clients.

And zero downtime in all 3 data centers.

In other words: business as usual.

We did update our phone systems to service our clients even faster.
Of course we are still reachable 24/7.

Our clients feedback will guide us along to optimize the details in implementation of this – like it does with everything else.

October 2012 was not very eventful.

Our New York data center went briefly offline due to Hurricane “Sandy”. All clients have been informed that
they do not need to pay their October invoice if they got impacted by this.

Utility power went offline and the backup generator failed.

We are overall very happy with the quality of our NY hosting vendors. We will evaluate carefully if we should update
matters on the east coast. By comparison INTERDUBS New York did behave fairly well.

Biggest “news” for September is that we adjusted our public client count to “more than three hundred”.
Here the list of companies that wanted to get listed publicly

In September everything was just humming along.

On August 21 larger uploads to our Los Angeles data center failed under certain conditions for a couple of hours. We have addressed the issue and eliminated the root cause of this bug in our system. Clients that have been impacted by this issue get credited based on their own impact evaluation. We promise 100%. And in case where we did not deliver entirely paying us becomes optional for our clients.

We added Highwinds as another Content Delivery Network option to our system.

Clients can choose now between:

INTERDUBS own network
Edgecast CDN
Highwinds CDN

for the delivery of their files. This can be done on a login by login basis – if specific situations require it.

All systems ran smoothly in July 2012.

The New York data center got adopted by the more clients and bandwidth exceeded expectations. During the migration we developed a couple of tools to make transition of data straight forward.

No downtime in June.

We upgraded our e1.interdubs.com data center to 6th generation hardware on June 30th. The software stack took a couple of days to stabilize. This impacted

* PDF icon creation
* iPhone clip auto transcoding
* plain vanilla FTP accounts
* network performance

Affected clients will get credited accordingly

We officially launched ny.interdubs.com to serve customers on the east coast out of 111 8th Avenue. Our INTERDUBS data center documentation has more details about this.

Clients using Time Warner Cable in Los Angeles had network performance issues between June 21 until June 25th using our west coast data center. While we started to make this known to TWC on the 21st it did take them a while to identify the equipment that caused the performance problem. We were able to route affected customers around by means of an alternative upstream provider.

INTERDUBS does not use Amazon or any other 3rd party cloud services. The Amazon outages on June 29 – 30 took allot of web services offline.

We were not impacted. We still have to make our own downtime. Not looking forward to it, but at least we will be able to learn from it and will make things work better in the future. With a 3rd party service all one can do is to switch to a new vendor. A bit like getting a new spouse when things don’t work. We believe in a approach to better things by working on the issues.

Zero downtime in May.

4,427 files short of four million.

We brought alternative upstream providers online. While we were and are optimally connected it can be helpful to have alternative routes on an entirely different IP range available. Bandwidth is a good thing. One can never have enough of a good thing.

April: zero downtime. One almost might want to add ‘of course’. What looks easy to the outside is still allot of work behind the scenes. That zero downtime is not normal might be revealed by a quick look at a service in the same space (we don’t like the word ‘competition’): Their April was impacted by 7 outages totaling 40 Minutes of downtime. A couple seconds more and they would have lost their third nine of uptime.

Our New York data center is still being tested. We rather take our time than to rush it into production.

In April we pushed 77 updates to the system. A good average of two improvements per day.

Speaking of numbers: We are looking forward to cross the 4,000,000 file mark in May.