We're building a credit card applicatoin, an advertisement serving platform, that will include many processes for example confirming, storing user data, demands, marketer info, pubisher info etc. Furthermore, we are producing optimisation files for everyone better advertisements.
We're using azure for cloud services. We're wishing to profit from appfabric on azure as well as their architecture.
My question relates to Azure SQL. Does anybody have knowledge about Azure DB, do you know the advantages, disadvantages? Performance, reliability, challenges?
While selecting Azure DB, what exactly are things must we benchmark or prototype?
Any ideas or thought, you want to share, especially associated with database on cloud? COnsistency, availability, Partial Tolerance, replication?
Benefit of SQL Azure is you can develop against quite a standard SQL database that individuals have plenty of experience with. Make migration of the application to Azure pretty easy.
The disadvantages i've come across are: 1. Limited backup. This really is improving soon, however it may not be the type of backup lots of people want. 2. Some SQL Server features unavailable, but 95% are. 3. No real treatments for database server hardware. 4. Cost may be an problem, based on your database size.
Apart from that, SQL Azure can be useful for me.
You will need to size your computer data.
SQL Azure is ideal for small (<1 GB or couple of GBs) with relational needs. It offers easy on-boarding for existing sql server designers and causes it to be quite simple to build up complex reviews.
However when the data size increases beyond a couple of GBs, it begins increasingly costly. Even the designs for scaling SQL Azure isn't as simple as scaling Home windows Azure Database - even though new feature of Federations should really allow it to be simpler, it's still different then scaling SQL Server (which individuals would normally scale-up - within SQL Azure there's a restriction of 50 GB and also the suggested pattern would be to scale-to more databases).
So, going through that which you have (demands, maybe user click data, etc) which could really increase a great deal, it may be easier to save all of this in Home windows Azure database rather than SQL Azure. You may also have a dual approach, by saving some data in SQL Azure to ensure that it's simpler to report against, after which have high quantity data held in Home windows Azure.
Architect well! It might save a great deal!
Even though this continues to be clarified already, I'd figure I add more info into it anyway.
Within the cloud, you've three storage options, each because of its own group of data:
- SQL Azure - here's your standard MS SQL database, with a few features presently disabled (note: they has stated, several occasions, each of the features is going to be switched on eventually). It guarantees reliability (AFAIK) by continuing to keep three copies of the data available at any time. Additionally you acquire some issues, like connection drops in some places, and you have to plan for your (within the repository logic, you must have retry logic). Clearly, just like other relational databases, chiefly helpful for storing relational data. Prices is alright as much as 1 GB (~ 10 USD) and rises significantly with as many as 10 GB (~ 100 USD) so make certain you really need it.
- Azure Table Storage - this can be a simple table storage. You primary secret is defined according to two IDs (Column &lifier Row secrets) and you will store inside it, pretty much, anything you like. You can't index the information (the answer posts are indexed, so that your queries could be pretty enhanced if guess what happens you do). This really is cheap (you have to pay by transaction), I believe the cost is about $,01 per 10.000 transactions. This can be used storage for a number of things, and you will also implement relations to some degree should you choose it right.
- Blob Storage - this can be used to keep your optimisation files. It is around $.15 per GB/month basically remember properly.
For performance factors, the blob storage is actually fast. I did not expect so that it is this fast. SQL does good too, it's in the end your normal MS SQL database. Becasue it is situated "close" for your instance, the network latency isn't that larger than it might be were it situated inside a separate machine inside your data center.
For current information on prices, see this Windows Azure Platform Offers. I am no expert in prices, so my figures are rough estimations.