When we adopted the 4*/5* naming convention in the mid-90s, I was a bit queasy 
about reserving so many addresses (2*4096) for CTCs, but we could afford it at 
the time. Since then, the number of DASD and tape devices has increased a lot. 

-- At the time, we managed two data centers with two independent IODFs, so 
duplicate addresses were possible and in fact existed. We later merged the data 
centers into a single IODF, which was a very good thing but increased the 
number of device addresses in use as duplicates were no longer tolerable. 

-- We implemented DR with a strategy that requires three copies per DASD 
volume: primary (production), secondary (XRC copy), and tertiary (flash copy of 
secondary for DR images). Again, a very good thing but a gobbler of device 
addresses.

-- Incremental increase in DASD usage over the years. Probably the smallest 
contributor to the increase, but as noted above, adding one volume to 
production is actually a three-fold increase in addresses. 

The advantage of any CTC naming scheme over just assigning random addresses is 
that a CTC can be fully identified by address alone: CEC, LPAR, and usage such 
as XCF vs. VTAM. The stories we heard about sysprog/operator confusion pretty 
much never materialized as the number of CTCs increased dramatically. 

.
.
J.O.Skip Robinson
Southern California Edison Company
Electric Dragon Team Paddler 
SHARE MVS Program Co-Manager
323-715-0595 Mobile
626-543-6132 Office ⇐=== NEW
[email protected]


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Ed Jaffe
Sent: Wednesday, October 17, 2018 9:40 PM
To: [email protected]
Subject: (External):Re: CTC conventions

On 10/17/2018 6:44 PM, Jesse 1 Robinson wrote:
> In hindsight I rue the decision to set aside both 4xxx and 5xxx addresses. In 
> practice, the low order digit is severely underused. In the beginning I 
> thought that we would need several different 'devices' for each LPAR. Never 
> actually found a need for more than four, 0 - 3. I like Rob Jackson's idea of 
> assigning even/odd numbers for gazinta and gazouta, or ranges like 0 - 7 and 
> 8 - F. Eliminating 5xxx, for example, would free up 4096 addresses for DASD.

It wasn't your fault. It was IBM's recommendation.

The fact is that all but one bit of position 'w' is wasted and position 'x' is 
nearly always underutilized in the scheme below. However, it's not uncommon at 
all to have more than 16 LPARs per processor making position 'y' too small in 
today's world.

It works great for small configurations, but should be reworked for larger 
environments -- perhaps by combining 'w' and 'x' to a single nybble and 
expanding 'y' to two nybbles.

CTC 4-DIGIT NAMING SCHEME: wxyz
w - one nibble to distinguish gazinta from gazouta x - one nibble that uniquely 
represents a processor; assigned by you arbitrarily y - one nibble that 
represents an LPAR on processor w, such as LPAR id from HCD z - one nibble to 
complete device address, assigned from 0 up to F

--
Phoenix Software International
Edward E. Jaffe
831 Parkview Drive North
El Segundo, CA 90245
https://www.phoenixsoftware.com/


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to