The answer to your question would depend upon the accrediting body you choose to use for your ISO 17025 accreditation. In general, you should choose an accreding body that is accepted worldwide. Based upon your profile, it would appear you are in a city in Mexico very close to the border of Texas.
Go to the following website:
http://www.ilac.org/documents/mra_signatories.pdf
This lists a group of accrediting bodies that all recognize each other worldwide. There are 5 accrediting bodies in the US and 1 in Mexico on this list. First of all, you should contact several of them (if not all) to get information on obtaining accreditation and compare them before choosing one to use for accreditation.
The simple answer to your question is that you need to use a calibration laboratory that is accredited by one of the organizations on this list. Make sure that the calibration laboratory has the measurement you need on their scope of accreditation. For virtually every parameter, there are several (or many) labs that will be accredited to calibrate your standards. Each laboratory will have a different uncertainty for their calibration. You need to choose the lab whose uncertainty is sufficient for your needs.
Of course, you can also send your standards directly to the national measurement institute directly. In the US, this would be NIST. NIST has signed a mutual recognition agreement with countries around the world, so you could also send your standards directly to those national measurment institues. A list of these is at
http://www.bipm.org/en/cipm-mra/participation/signatories.html
Sending your standards directly to NIST (or similar) gets you a lower uncertainty on the calibration. Calibration by NIST is very time consuming (months) and extremely expensive. The lower uncertainty you get will be statistically insignificant for routine calibration work.
For example, if you send gage blocks to NIST, the uncertainty might be 1 microinch. If you send them to Starret, you might get 2 microinches. If you send them to any number of accredited labs, you would get between 4 and 8 microinches. When you use the gage blocks to calibrate a micrometer with 0.0001 inches, the uncertaty from that calibration is roughly 58 microinches regardless of where you sent the standard.
I hope this helps a little.