Database Reference
In-Depth Information
100000
10000
Write to Azure Blob Storage
Write to Amazon S3
Read from Azure Blob Storage
Read from Amazon S3
Delete from Azure Blob Storage
Delete from Amazon S3
1000
100
1 MB
10 MB
15 MB
Data Size
Fig. 4.12
The database processing time of read, write, and delete in cloud databases with low
stress large file read, write, and delete test sets
in Amazon EC2 now performs the worst among all platforms. It implies the
poor capability of handling concurrent requests within the same instance as the
compute capability. Moreover, Google App Engine Datastore, Amazon SimpleDB
and Microsoft Windows Azure Storage all continue to show faster speeds in read
operations than write operations.
Low Stress Large File Read, Write, and Delete
Figure
4.12
shows the average database processing time of reading, writing and
deleting binary files in the cloud databases directly. It can be seen that reading,
shown in the left figure, is faster than writing, shown in the middle figure, in general.
Both database processing time of read and write for Amazon S3 and Microsoft
Windows Azure Blob Storage are linearly increasing with increasing proportion of
data size. It is likely the limitation of the local network environment will come
before getting insights of the cloud databases. This is why the CARE framework
provides a range of scenarios, for example, end-user-cloud database, as well as cloud
host-cloud database, so that the performance characteristics can be evaluated with
and without the network variations and effects in place.
The average database processing time of the delete operation, shown in the right
figure, is interesting as the observation shows a constant result regardless of data
sizes. It is confirmed that neither Amazon S3 nor Microsoft Windows Azure Blob
