This post is more than 5 years old
1 Rookie
•
15 Posts
0
4460
December 14th, 2008 16:00
'Single Instance' is based on Blob or Clip?
If I store the same file into different Tags in the same Clip, the Content Address will be the same for the two Tags? but when I checked out the Raw data in the Clip, I got the following message:
the md5s are different. Did I misunderstand something? md5 is not the Content Address? or the 'Single Instance' is based on Clip, but I tried still no same md5s.
No Events found!
gstuartemc
2 Intern
•
417 Posts
0
December 16th, 2008 04:00
The cluster Strategy is set by the Centera administrator (for the online clusters, that's us). In Performance mode, all blobs smaller than 250K are written with a GUID as part of the content address to facilitate co-location performance gains.
If you set Collision Avoidance to be TRUE, it is because you never want the miniscule chance of a collision and in this case a GUID is added to the Content Address to esnure the uniqueness of the blob. So if you want Single Instancing, then Collision Avoidance must be set to FALSE.
The Online Clusters are all set for Performance stratgey - if you want to test Single Instancing, ensure you work with objects larger then 250K.
M indicates the the Content Address is calculated using the MD5 algorithm. M++ indicates it is calulated using a hybrid MD5 / SHA256 algorithm.
gstuartemc
2 Intern
•
417 Posts
1
December 15th, 2008 07:00
Single Instance is based at the Blob level.
I would prseume that you are wring the content to a cluster that is set to Storage Strategy Performance, or that your application is setting Collision Avoidance to true (which adds a GUID element).
The M and M++ naming schemes allow for Single Instance Storage using Storage Strategy Capacity on the cluster.
disorderdev
1 Rookie
•
15 Posts
0
December 15th, 2008 18:00
Thanks very much!
I'm a newbie here. so It seems that I did not config Centera properly. but I got some new problem:
1. how to set cluster Storage Strategy Performance?
2. if I set the cluster Storage Strategy Performance or Collision Avoidance to true, then the md5 will not be the same? but I didnot change any online cluster's configuration.
3. what does the sentence "M and M++ naming schemes" stands for?
BTW: I'm using EMC online cluster, could I use it to test single instance?