Hello,
We have got an applications where we query about 50 types of snmp OIDs. The types of OIDs are drops into several types such as integer, double, date, string and float.

We are planning a structure to keep type information in a different table and keep the data in single column in a different table. Our aim is to build an architecture flexible to be able to add new OIDs in future.

Some of the OID needs to be pulled a lot more than the others.

Our question is what will be the performance figures as oppose solution. The oppose solution would be to have each OID in a table column with its type known.

I hope description is enough for you to see the problem.

Regards

what is an snmp OID?

how can it have different values? it's either numeric or it's a string...

Hi,
Assume they are name of variables such as

1.2.4.1.1.2 is an integer
1.2.4.1.1.4 is an double
1.2.4.1.1.5 is and string

we need to put all these variables in a column and put their type to another table. While retrieving them use a join with types column and data column.

The other alternative would be to put all variables to a different column but when the number of OIDs increase management becomes a problem.

We are just wondering if this type of mechanism cause trouble in long term as number of data increase.

what? that doesn't make any sense

those are strings, full stop


huh? need to?

Hi,
Its very common, SNMP, may be need to look at a SNMP tutorial at google.


1.2.4.1.1.2 is an integer
1.2.4.1.1.4 is an double
1.2.4.1.1.5 is and string

1.2.4.1.1.2 this is variable's name, it has a integer type and its value will be placed in to database.

its just like int i = 50 ; int is type , i is name of variable and 50 is value
1.2.4.1.1.2 (int) = 50

My question is conceptual. Don't couple your self with OIDs.

Question is if you have 50 different variables and has to be put in a database and we expect it to be flexible for future additions of new variables.

Search for the (anti) pattern "Entity Attribute Value"

Im guessing 1.2.4.1.1.2 is an IPv6 address
the value is the response or reading for that IP address
however what is missing in your schema is the OID itself and a datestamp indicating when the event was logged
id expect
IPaddress
OID
EventTime
Response

I guess it depends on what you want to do with the information. if its purely for logging purposes then you could log the response as a string and coerce it back to numeric at a later date if required. all though it runs fingernails down the blackboard of my soul to suggest such things.

if however you need to analyse the response then you may need to thin of redesigning the response to a super/subtype.