Last year we got 5" of snow. This year....~60"+. Between all the "polar vortexes" (who thinks up these things??? what happened to the good ol' "Arctic Blast"...), I think brain freeze has settled in. Thankfully, Mike and the others at ISUG decided to host the ISUG Tech conference in Atlanta in April. YEEESSSSSSSSSSSSSS!!!! Maybe, just maybe, when I come back the snow mounds beside the drive and in the front yard will be gone. It got so bad this year that both of my Alaskan Malamutes became Alaskan Lodge Dogs.
Years and years and years ago, I spent a lot of time in Atlanta working on a project for then-Sybase. I can say, April is nice in Atlanta.....tons nicer than Orlando in August where you have to swallow to breathe. For those of you coming, I have some suggestions:
1) Don't arrive during rush hour.
2) If you are used to driving the Jersey Pike in non-rush hour when it is moving 80mph bumper to bumper - you will be right at home in Atlanta. This is the south....they love NASCAR...it is called bumper drafting. If you are coming from CA where traffic is perpetually gridlocked, arrive during rush hour and you will feel right at home.
3) Come early...and spend the week.
....okay, that last one was a subtle hint for a shameless plug. Thanks to ISUG (and whomever was selecting sessions), I got a nice bodacious 8 hour timeslot to teach RS Internals and P&T. Yes, we will cover RS MC....and try to show you how it all maps back to the threads/modules/message queues and other configuration parameters. Unfortunately (or Fortunately?), it will be on Monday. That's the day before the conference itself starts. At least it won't be on Friday like my colleague Chris Brown's MDA session (although I did promise to help him with that). At least the folks who attend my session will be wide awake and not dreaming of chocolate bunnies, marshmallow peeps and colored eggs. Hmmmmm....there's and entertaining idea for Friday's session....an egg hunt. Am debating on getting a bunch of thumb drives, putting a VM on it (Linux, of course) and sending students home with it......but I need to figure out how to do so economically as SAP has all these funny rules and policies (aka red tape).....and I likely am going to have to sneak it past the wife's budget filter and her dreams of a kitchen makeover.
In addition to the RS P&T session, I am doing a session on ASE statement cache tuning.....because if you don't get it right, you can peg CPU through the roof as I found out the hard way. Another session on ASE VLDB management - a class I hope to make interactive as we will be talking dbcc's, reorgs, create index and all the other pain points of multi-TB systems....and some tricks I have learned. Stefan Karlsson will be there talking about his favorite topic of spinlocks..... oh, yeah, this is a SRS blog......
Sooooo....one of the sessions - most likely taught by me - will focus on RepAgent peformance. Those of you who attend Monday's session will be able to spend that time somewhere else as there is a bit of overlap. However, if you can't get away from work on Monday or prefer to spend Monday in Joe Woodhouse's sp_sysmon session (now I know why Mike scheduled me for Monday...he didn't want me in there heckling Joe......just kidding, of course)....then I will be happy to explain to you why RepAgent always was slow in ASE 12.x and early 15.x and what you can do about it in 15.7+...as well as talking about multi-threaded scanners....
Speaking of RepAgent.....one of the topics there will be discussing Synchronous RepAgent. While it will support zero data loss, it becomes more interesting when viewed in the broader context. ASE 16 will be introducing a whole new feature set called ASE HADR. Of course, this is the first release and so all eventual functionality is not in it yet....but I am starting to see the glimmers of a pathway to official "virtual clusters" - an area I think SAP could differentiate ASE from others and meet a huge gap in the market that many many customers have cobbled together often frightening solutions around. Regardless, the two together solves a major headache a lot of customers have had using standby replication.
One of my other "likely" sessions will be on direct load. Now THAT is a feature I love. To me, you can take all the other create subscription thingys and just toss them (except without materialization). You just need to know two create subscription commands - without materialization and direct load. All those years of trying to keep atomic, sub-atomic, bulk and whatever-else-subscription method can just be tossed. I only see ONE littttttllllleee problem with it. It doesn't support SAP IQ. ....although maybe with IQ 16.......hmmmmmmm..... It *does* support ASE targets though......and I see lots of opportunity here - especially as the speed improves.
My good friend Cory Sane from NC is making the short hop down there as well....and talking about how his <insert major bank> used SRS to move HW platforms with minimal downtime. .....and Phil Adams will there talking about using SRS to feed IQ via RS/RTL. Now, I have to admit, I met Phil at a customer site where he was working...on a common <insert health care package here> installation and I truly think that the things that were done there likely could be replicated in a lot of other places....I know of at least 3 other customers that should take seats in his session....assuming they attend. Of course, if they don't, Phil gets to peddle his expertise to them and charge exorbitant rates.
Over all, it should be a bang-up conference...as usual, I hate to be teaching as there are a lot of sessions I would like to go to as well. Guess I will just have to download them later....sigh.....
