Repository
Repositories are tightly coupled with proxy servers or a proxy pool. While a single proxy server or proxy pool can connect with multiple repositories, each repository is uniquely coupled with a single proxy server or proxy pool.
Please also refer to the Proxy section for more details.
Repositories can be based either on Object Storage or on the local Microsoft Jet database (.adb) when used in combination with standalone proxies. When using a proxy pool, Object Storage is the only option. There are significant advantages to using Object Storage, as it offers more efficient compression, as well as support for encryption and immutability. Local Microsoft Jet database repositories can be stored on Direct Attached Storage (DAS) or Storage Area Network (SAN) volumes.
General Best Practices
VCSP and Enterprise setups should always aim for the usage of Object Storage Repositories
Use Object Storage Repositories whenever possible - either on-prem or cloud - as it offers more efficient compression, as well and support for encryption and immutability.
If using local Microsoft Jet databases, avoid very large repositories by separating repositories by data type (Exchange, OneDrive, SharePoint) and using multiple repositories per data type when required
You want to avoid very large repositories because handling these becomes complicated in case of migrations and rescans. Larger repositories can be avoided by firstly splitting based on data types but also by splitting the same data type, e.g. based on proxy alignment or business units. Splitting by data type makes additional sense, because characteristics for each data type (Exchange, OneDrive, SharePoint) is different and thus it can be handled different in separate repositories.
We consider 200-300 TB as managable size.
It has been observed that large local repositories based on Jet databases tend to get slower by the end of the year until new database files are created with the change to the next year.