Update 8th July – the plot thickens
After posting this blog two days ago, it seems Oracle Corp. may be taking a little time out to “clarify” their position. The original MOS note referenced below is no longer available (despite the link still sitting in my favourites in MOS), and furthermore, the helpful and rather detailed blog post by Mike Dietrich (News on Oracle Database STANDARD EDITION 184.108.40.206) has been redacted to a “watch this space” note.
So I guess we’ll have to watch this space…
My original blog posting from 6th July remains below.
Disclaimer – these are my personal musings based on very limited information, which I’m sure will rapidly become out of date as more details emerge over the coming months.
After a year of uncertainty, Oracle has finally announced the future of Standard Edition (SE). Along with its smaller sibling Standard Edition 1 (SE1), it is being retired and replaced by Standard Edition 2 (SE2). Perhaps the most significant point of this change is that SE2 can only be run on one- or two-socket servers, which raises interesting questions for existing SE deployments on four-socket servers.
When 220.127.116.11 was released in July 2014, it was only available for Enterprise Edition (EE). No date was given for SE and SE1, and with patching for 18.104.22.168 scheduled to end in July 2015, there was an uncertain future for SE and SE1. With Extended Support fees waived on 11g for a year, it seemed like sticking with 11g was a better option for SE and SE1 than moving to 12c with no clear future. However, with 11g ending Premier Support in early 2015, things were beginning to look very bleak for Oracle’s most cost-effective database platforms.
After months of silence, a MOS note was quietly published in early July 2015 (2027072.1) with a rather surprise announcement. The key points I took from this brief note are :
- 22.214.171.124 is the final release for SE and SE1
- Beginning with 126.96.36.199, SE and SE1 will be superseded by SE2
- SE2 can be run on a maximum of 2 sockets, and can optionally support a two-node RAC cluster
- Customers running SE or SE1 will have to migrate their licenses to SE2 if they wish to upgrade to 188.8.131.52 (i.e. if they wish to remain supported beyond January 2016)
- SE2 is scheduled for release in Q3 2015
- Patching for 184.108.40.206 SE and SE1 will be extended until January 2016, beyond the originally published date of July 2015
So what does this mean for SE and SE1 users?
Firstly, there’s a small lifeline, in that 220.127.116.11 will be patched for another six months. However, if like me you’ve stuck with 11g and haven’t yet upgraded to 12c, is it really worth upgrading for a further six months, when 11g remains in (fees-waived) Extended Support for the same period?
The most interesting point is what happens to systems running SE on a 4-socket system, or a 2×2-socket RAC system. When the lights finally go out on SE and SE1 in January 2016, you will have to migrate to different hardware to keep within the 2-socket limit. Alternatively you may be faced with a large bill to upgrade to EE.
There’s been no mention of pricing yet. It’s been suggested that the cost of SE2 could fall somewhere between SE1 and SE. However, with the MOS note saying you have to migrate your licenses in order to upgrade, I wonder how many customers will end up paying Oracle for the privilege of migrating. Cue some lengthy discussions with your favourite account manager.
What will SE2 look like? Will it keep to the same limited set of features as SE/SE1, or will a few EE features or options be allowed to creep in? Parallel execution would be a nice treat. Will SE2 offer any improved performance monitoring capabilities, or will we still be forced to hunt around for the install script for Statspack?
Finally, will Oracle use this as an opportunity to retire the per-socket metric for licensing of SE/SE1, instead moving to per-core licensing, as is done with EE? With the advance of multi-core chips, it seems inevitable at some point. Again, this could adversely impact customers who have invested in high-end multi-core chips.
I’m no lawyer, but having recently been debating perpetual vs term licenses, it makes me wonder what “Perpetual” really means, when support for said license is no longer available. I’m sure it’s in the small print somewhere.