Hi All,
I have some doubts regarding Hashed File. In Datastage Ver4, Key column is a must in column definition otherwise it will give the message box saying "Key column should be selected" and it wont allow the users to press OK button without key column. But in DataStage Ver 5 and V6, it is not explicitily asking for the Key column. Even if there is no key column, it is accepting. Can anyone explain abt this in detail???
One more thing, I want to know the basic difference between Universe Table and Hashed File.
Also, what is difference between key column in between these two??? Because, I have created a Universe table and I am able to access the same table thro hashed file stage. So, I want to know the differenc ein detail. Can anyone throw light on this???
Thanks in advance,
Bala
Hashed File
Moderators: chulett, rschirm, roy
-
ray.wurlod
- Participant
- Posts: 54607
- Joined: Wed Oct 23, 2002 10:52 pm
- Location: Sydney, Australia
- Contact:
A hashed file is the way in which the UniVerse database implements its tables. If you create a UniVerse table, you get a hashed file plus system table entries. This explains why you can access a hashed file via a UV stage and a UniVerse table via a Hashed File stage.
As you probably know, the DataStage Engine has begun to diverge from UniVerse since release 5.0, but the differences are not all that great thus far - mainly removing things that UniVerse needs but which DataStage does not need (for example transaction logging, replication)
Hashed files are implemented using a computational mechanism (called "hashing") for determining storage location of key values. Among other things, this means that hashed files (UniVerse tables) do not need an index on the primary key.
Accordingly, a hashed file can NOT be created without at least one column being defined as the key value.
You will find that, if you create a table definition that lacks a key column, DataStage will assume that the first-named column is the key column, even switching it to NOT NULL if required. OK, this is a change in behaviour from release 4.0 but, as noted earlier, release 5.0 allowed Ascential the opportunity to "do their own thing" without having to retrofit to UniVerse behaviour.
As you probably know, the DataStage Engine has begun to diverge from UniVerse since release 5.0, but the differences are not all that great thus far - mainly removing things that UniVerse needs but which DataStage does not need (for example transaction logging, replication)
Hashed files are implemented using a computational mechanism (called "hashing") for determining storage location of key values. Among other things, this means that hashed files (UniVerse tables) do not need an index on the primary key.
Accordingly, a hashed file can NOT be created without at least one column being defined as the key value.
You will find that, if you create a table definition that lacks a key column, DataStage will assume that the first-named column is the key column, even switching it to NOT NULL if required. OK, this is a change in behaviour from release 4.0 but, as noted earlier, release 5.0 allowed Ascential the opportunity to "do their own thing" without having to retrofit to UniVerse behaviour.
-
ray.wurlod
- Participant
- Posts: 54607
- Joined: Wed Oct 23, 2002 10:52 pm
- Location: Sydney, Australia
- Contact:
