When your data is put out into the Cloud, you rely on the schemas and data structures of your software as a service (SaaS) application vendor to provide you data in a way that is useful to you
2. When your data is put out into the Cloud, you rely on the schemas and
data structures of your software as a service (SaaS) application vendor to
provide you data in a way that is useful to you. Sometimes that works and
sometimes it does not. Often, you need to have special reorganizations or
views of those data elements.
A trivial example would be a table in a billing system that is invoices. You
get that invoice data and bring it into a spreadsheet, and you start doing
the math on how much people owe you based on these invoices by
month, but you do not realize you have brought over all the invoices.
There are cancelled invoices, invoices that are on credit hold and invoices
that have not been sent yet, but what you would rather have is a virtual
view – or several virtual views – of that invoice data.
3. You want one that includes all the invoices, one that shows the active
invoices and one that shows the invoices that have been sent or are
overdue. You probably want to virtualize that data, and that is particularly
important when you have something like revenue recognition that might
involve four or five tables. Some of those tables might not even be in the
billing system. Some of them might be in the customer relationship
management system. You may have some information in the accounting
system as well. You have to bring all this information together, and in a
complex set of queries, or query script if you will, that creates a bunch of
temporary tables, then rejoins data and reunions them together to give you
the transactions that are in the recognizable revenue category.
4. It is important for users to be able to see these transactions as tables. A
user should not have to worry about how they have to get rid of the invoices
that are not important. They should not have to worry about the mechanics
of how to go across three applications and five different tables to get the
recognizable revenue. They should just see things that look like a table, and
they are going to want to do their own selection projection and joining on
that table. They are going to want to limit it to just last week or last month, or
compare it to the same quarter last year. They will have their own
manipulations they will want to do, but it is important that data be virtualized.
5. The users want to be able to not only query the raw data, but also be
able to get at very usable views of that data, or virtual tables if you will.
I think people are thinking this way a lot more than they used to now
that we have virtual machines. A virtual machine, of course, virtualizes
the central processing unit (CPU), memory, disk drives and network
connections. It is time for us to start thinking that way in terms of the
most important thing we do in data processing, and that is dealing with
our data.