This is surprising and I'm not sure how to feel about it. It did seem that WCG was running out of steam - so to speak - as projects were either on pause or completed. Krembil shouldn't have any lack of work, if they have the resources to take advantage. We'll just have to wait and see what happens.
I've been taking a back seat on computing for a while, enjoying just letting the devices run with minimal management. Figured I would pop back in on this one, though. I hope everyone is doing well.
My first reaction was annoyance that IBM would drop WCG while it continues to make billions in net income per year (i.e., free and clear profit after all expenses and taxes). My guess is that they realized that they could 1. Save/redeploy the staff costs, 2. Save the IBM Cloud use costs, and 3. Most significantly, redeploy the IBM Cloud assets WCG was using to other uses. As IBM Cloud was, I believe, their most profitable sector, they probably want to make the best financial use of it they can. I'm also really disappointed that IBM didn't find new roles for the entire WCG team.
My second reaction was that this could be a great shift to allow more projects onto the grid. I think we've all experienced the frustration of the long on-boarding times with WCG in the past, partly from needing to code new science applications (which clearly won't be solved by the move) but also from the IBM security and legal teams. While I appreciate the thorough security reviews that IBM did, it would be nice if we could get more, small projects on WCG that rotate through instead of it needing to be a huge, multi-year project to make it worthwhile to add. I'd like to see a nice group of long-running projects that always have work to keep the grid busy (Mapping Cancer Markers, Open Pandemics, etc.) while we then have targeted projects that last a couple of months doing critical work that would have been too small for WCG in the past.
And finally, I do have concerns on the technical side. WCG was one of the few (maybe the only) BOINC project that seemed to always be up and running. Downtime was minimal and mostly due to security patching. Whenever there was an issue, the tech team was on it. I hope WCG doesn't turn into other BOINC projects with unexpected downtime of days or weeks and little to no communication from project admins or techs.
I'm not at all surprised that IBM dumped WCG. I'm surprised they kept at it for 17 years. It was a philanthropic thing to begin with so I figured it might last 5 years and then IBM would find other places to put that money. It is rare to have a company (especially one like IBM) continue to support an endeavor that isn't bringing in any revenue. On the balance sheet it is an expense and that has to be answered for to investors. My son retired from IBM July 1 at the age of 40. IBM is going through some changes right now with spinning off some of their services groups(Kyndryl is the new name for IBM's managed infrastructure services business, which will spin off as a separate company by the end of 2021.) and are trying to make the IBM parent company look as good as possible. Additionally, the WCG team has been pretty much intact since the project started and most were IBM employees before WCG started. Most were probably approaching retirement in the next few years which meant IBM would have to find replacements now so they would be trained by the time the others left. IBM probably decided to relinquish the project instead of going through that process. In other words, this would be a good separating point.
I'm mixed on the move to a research organization, especially a non-profit. Supporting this infrastructure at the IBM scale is going to cost a bundle. What really costs the most money are the storage units and supporting infrastructure (fibre switches and redundancy). You can spend $100,000 on servers but $1,000,000 on storage. My understanding was, IBM had Terabytes of storage on the floor at their hosting site. IBM also had access to hardware through their own organization and had access to technical resources (Hardware engineers, software engineers, open-source engineers through their RedHat organization etc). Where is this new organization going to get that? On the other hand, because it was IBM, everything moved at a snail's pace. Very secretive about any new announcements or upcoming additions. Maybe the new organization will be more open and innovative with the project. It is going to be interesting to see how the transfer happens. Based on what I have read, it doesn't sound like the hardware is being transferred so there will probably be downtime to switch over to the new hardware. The big thing is going to be moving the data. WCG will have to stop at some point, move the data ( which could take days) and restart. Now they could, depending on resources (people and money) setup synchronous storage replication (depending on distance) and then just stop the IBM side when the time comes. Less downtime but WAYYYYY more costly and technical. Stay tuned....
I've taken the stance, that I'm going to see how it goes and hope for the best. I've been frustrated by the way IBM has managed the project in the past and have moved away from it over the past few months. I'm in the process of moving to a new location and will be limited in contributions until I can get a new home built. Once that happens, I have decided to rebuild my farm and start contributing on a greater scale again. I may rejoin back to WCG and use this period of reduced contributions to "test-drive" the new organization. It may actually turn out to be very good thing...