Archive for July, 2011
Couple of days ago I had to explain a colleague at the customer side some things about Windows Azure. And of course Scalability comes around the corner. I saw myself drawing all kinds of pictures to make it clear. So I made a nice presentation, here are a few bits of it.
At a Windows Azure customer of ours they keep track of changes on the tables in the SQL Azure database. This is done by a trigger, the trigger inserts the change in a table. I can not think of another solution without losing data or changes.
As told in my previous Blogpost I kept a list of the Blob name and the LastModifiedUtc. To make this solution usable for more then one instances I stored this list in a Windows Azure storage table.
I made a solution for a customer where I copied blobs from Windows Azure Blobstorage to an different location on Windows Azure Blobstorage. But to be a bit efficient I was keeping a list of blobs I have copied already. This list contains a Last modified date. So I the Last modified date was not changed I did not copy etc.
Speed-up of connection
Having a fiber connection for TV and radio, next phase is Internet. The basic offering is 50Mb – symmetrical, meaning a 6-fold increase of download, but far more interesting for the reader and user: a 50-fold increment of upload speed. It means I had to change DNS settings, as I learned that the address I got from the new ISP is functionally static. That has been tested by connecting my laptop directly to the moden, and afterwards my router – after keeping the connection open for about one hour. I got the right address! The router I use has two WAN ports, so it’s an easy transfer: connecting the fiber into the second WAN port, and supply the ‘old’ address as an alias.
Today I got the acknowledgement of the change – indeed, when accessing the websites I got the new address, but no answer…Luckily, I could access the router and found out I had to disable the first WAN port since that would supply the now obsolete default route; Second, I needed to add a set of open pofrts to the second WAN port. Actually, I made the first set available on the second WAN port and copied it to be used on the first. Now I can access the webs (therefore I cab write this post!), mail gets in and I can access the server.
The difference is speed is noticable, so say the least. Even on a Wifi connection with poor signal
In my previous post I described how the IO resource manager (IORM) in the exadata storage can be used both to guarantee a minimum amount of IO a database can get (which is what is covered in most material available), but also to set a maximum amount of IO. This is what Oracle calls an inter-database resource manager plan. This is set and configured at the cell level using the cellcli with ‘alter iormplan dbplan’.
There are different ways on the Windows Azure platform to make and set configurations.
For the WebRole we have the web.config and for the WorkerRole we have the app.config. These config files are not new and we are used to use these. The advantage of this is, porting our current applications to the Windows Azure platform do not need many adjustments. Disadvantage these settings are only changeable with a new deploy of the application. In principal with RDP to the Role, you are able to change the settings in the web of app.config . But these changes will disappear during a reimage of the instance and are not valid for all the instances of the role.
The Exadata IO Resource Manager (IORM) is a fundamental piece of the exadata database machine since the beginning. The function of the IORM is to guarantee a database and/or category (set via the database resource manager) gets the share of IO as defined in the resource plan at each storage server. The “share” here means the minimal amount of IO. This is widely known and documented in the Exadata documentation. Having a minimal amount of IO means you can guarantee a service level agreement (SLA).
Yes, finally. There has been an update on the Windows Azure Management portal. No new functionality was added (however I have some whishes left), but the UX of the Management portal is more intuitive. Before when you accidently did a Right Mouse click, you got the vague message: it is a Silverlight site. Yeah so what . Luckily of course this means Silverlight is far from dead (but that is a different discussion), but the vague message was not very functional.
These last weeks I was notified that new user acounts were created on the OpenVMS wiki. Looking at these users, I had my suspicion but I gave them the benefit of doubt. It meant I had to watch activity on the wiki, on a daily basis. But my thoughts proved to be right: two of three had created pages that have nothing to do with the subject – merely hawking stuff. So I deleted the pages – and the users. So these won’t be accessable anym more, still have to move them off the Python environment.