More often than not, database administrators around the technologies have to fight out high load on their databases. It could be ad hoc queries, urgent reports, overflown jobs, or simply high frequency and volume of queries by the end users.
DBAs try their best to do a generous capacity planning to ensure optimal response time and throughput for the end users. There are various scenarios where it becomes very hard to predict the demand. Storage and processing needs in case of unpredictable load are hard to foretell in advance.
Cloud computing offers the promise of unmatched scalability for processing and storage needs. Amazon AWS has introduced a new service which gets closer to that ultimate scalability. Amazon Aurora is hosted relational database service by AWS. You set your instance size and storage need while setting Aurora up. If your processing requirements change, you change your instance size and if you need more read throughput you add more read replicas.
But that is good for the loads we know about and can more ore less predict. What about the loads which appear out of blue? May be for a blogging site, where some post has suddenly gone viral and it has started getting million of views instead of hundreds? And then the traffic disappears after some time suddenly just like it appeared out of nowhere and may be after some days it happens for some another post?
In this case if you are running Amazon Aurora, it would be fairly expensive to just increase the instance size or read replicas in the anticipation that some traffic burst would come. It might not, but then it might.
In front of this uncertainty, enters Amazon Aurora Serverless. With this Serverless Aurora, you don't select your instance size. You simply specify an endpoint and then all the queries are routed to that endpoint. Behind that endpoint lies a a warm proxy fleet of database capacity which can scale as per your requirements within 5 seconds.
It's all on-demand and ideal for transient spiky loads. What's more sweet is that billing is on second by second basis and deals in Aurora capacity units and minimum is 1-minute for each newly address resource.
DBAs try their best to do a generous capacity planning to ensure optimal response time and throughput for the end users. There are various scenarios where it becomes very hard to predict the demand. Storage and processing needs in case of unpredictable load are hard to foretell in advance.
Cloud computing offers the promise of unmatched scalability for processing and storage needs. Amazon AWS has introduced a new service which gets closer to that ultimate scalability. Amazon Aurora is hosted relational database service by AWS. You set your instance size and storage need while setting Aurora up. If your processing requirements change, you change your instance size and if you need more read throughput you add more read replicas.
But that is good for the loads we know about and can more ore less predict. What about the loads which appear out of blue? May be for a blogging site, where some post has suddenly gone viral and it has started getting million of views instead of hundreds? And then the traffic disappears after some time suddenly just like it appeared out of nowhere and may be after some days it happens for some another post?
In this case if you are running Amazon Aurora, it would be fairly expensive to just increase the instance size or read replicas in the anticipation that some traffic burst would come. It might not, but then it might.
In front of this uncertainty, enters Amazon Aurora Serverless. With this Serverless Aurora, you don't select your instance size. You simply specify an endpoint and then all the queries are routed to that endpoint. Behind that endpoint lies a a warm proxy fleet of database capacity which can scale as per your requirements within 5 seconds.
It's all on-demand and ideal for transient spiky loads. What's more sweet is that billing is on second by second basis and deals in Aurora capacity units and minimum is 1-minute for each newly address resource.
No comments:
Post a Comment