You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If our current use of traits doesn’t hit the spot, maybe we can consider this approach…
If I image a rack of devices all doing the same function (e.g. SDI to 2110-???), in terms of configuration there is some settings that all units will have (e.g. codec type), some that is specific to the device (e.g. device name, serial number) and some that is specific to the streams that are being sent/received (e.g. IP addresses, lost packet accounts, certain tags). That suggests to me that there are three categories of information: Common settings (let’s call them General), Device Specific (let’s call them Device) and Stream specific (let’s call them Stream).
Then there is stuff that a controller can change and that which a controller can’t change. Let’s call these Writable and Read-Only.
Combining these, we get categories of Device Writable(e.g. device name), Device Read-Only (e.g. serial number), General Writable (e.g. SIPS/FEC settings) , General Read-Only(e.g. IS-04 UUID?), Stream Writable (e.g. IP addresses) and Stream Read-Only(e.g. lost packet counts).
A Backup Data Set would include properties with all six categories, but when restoring a Data Set a user would specify one or more of Device, General, Stream. Properties with a Read-Only category would be ignored when restoring and no warning would be generated.
A full kit replacement restore would select all three categories.
A kit parking or change of program would likely only select Stream.
A mass-provisioning via cloning would initially just select General(or possibly also Stream with some default values).
Jeff’s suggestion of rebuilds vs. restores plays in well to this. To make it easier for me, building on the terminology of C++ compilers, how do people feel about calling these a clean restore and an incremental restore? These could be flagged as part of the restore operation. (A mass-provisioning could do a clean restore with just the General properties so wouldn’t technically be a rebuild.)
The text was updated successfully, but these errors were encountered:
If our current use of traits doesn’t hit the spot, maybe we can consider this approach…
If I image a rack of devices all doing the same function (e.g. SDI to 2110-???), in terms of configuration there is some settings that all units will have (e.g. codec type), some that is specific to the device (e.g. device name, serial number) and some that is specific to the streams that are being sent/received (e.g. IP addresses, lost packet accounts, certain tags). That suggests to me that there are three categories of information: Common settings (let’s call them
General
), Device Specific (let’s call themDevice
) and Stream specific (let’s call themStream
).Then there is stuff that a controller can change and that which a controller can’t change. Let’s call these
Writable
andRead-Only
.Combining these, we get categories of
Device Writable
(e.g. device name),Device Read-Only
(e.g. serial number),General Writable
(e.g. SIPS/FEC settings) ,General Read-Only
(e.g. IS-04 UUID?),Stream Writable
(e.g. IP addresses) andStream Read-Only
(e.g. lost packet counts).A Backup Data Set would include properties with all six categories, but when restoring a Data Set a user would specify one or more of
Device
,General
,Stream
. Properties with a Read-Only category would be ignored when restoring and no warning would be generated.A full kit replacement restore would select all three categories.
A kit parking or change of program would likely only select
Stream
.A mass-provisioning via cloning would initially just select
General
(or possibly alsoStream
with some default values).Jeff’s suggestion of rebuilds vs. restores plays in well to this. To make it easier for me, building on the terminology of C++ compilers, how do people feel about calling these a
clean restore
and anincremental restore
? These could be flagged as part of the restore operation. (A mass-provisioning could do aclean restore
with just theGeneral
properties so wouldn’t technically be a rebuild.)The text was updated successfully, but these errors were encountered: