Microsoft Exchange Server 2019
At the Ignite 2017 conference Microsoft announced their plans for the next version of on-premises Exchange Server, and they're calling it Exchange Server 2019.
A preview of Exchange 2019 will be available some time in mid-2018, and the release is targeted for the second half of 2018. This follows the typical release cadence for Exchange Server every three years, with RTM of a new version slightly before the year that it's branded with. Exchange 2016 was released in late 2015, Exchange 2013 was released in late 2012, and Exchange 2010 was released in late 2009.
Development for on-premises Exchange has followed a pattern of new features arriving in Exchange Online first, and then making its way to on-premises customers via cumulative updates or as an entirely new release (e.g. some cloud improvements in the Exchange 2013 era did not arrive on-premises until the 2016 release).
Although Microsoft is not making any specific announcements at this stage, that doesn't stop us from speculating as to what we'll see in Exchange 2019. My predictions for Exchange Server 2019 are based on the areas Microsoft called out in their slide shown above, some assumptions, previous patterns, and the feasibility of bringing cloud improvements to on-premises infrastructure. As the joke goes, Exchange 2016 was more like Exchange 2013 Service Pack 2, so perhaps Exchange 2019 will look on the surface a lot like Exchange 2013 Service Pack 3 when it arrives.
At this stage (and I stress, these are predictions only) I'm expecting the following from Exchange 2019:
Minimum requirements for operating system and Active Directory will exclude any that are out of mainstream support. This would make Windows Server 2012 R2 the minimum OS and AD for Exchange 2019, if Exchange RTMs before October 2018 when Windows 2012 R2 goes into extended support. If Exchange 2019 RTMs later, we may see the minimum requirements set to Windows Server 2016 instead.
Features that were deprecated in Exchange 2016 will be removed entirely in Exchange 2019. This means the end of RPC-over-HTTP, which has been replaced by the improved MAPIhttp protocol instead. Since MAPIhttp is supported by all Outlook clients that are in mainstream or extended support today, there's no need to retain RPC-over-HTTP. Server-side, this could also mean the removal of “Outlook Anywhere” as a configurable Client Access namespace, with the MAPI virtual directory taking over the job.
Outlook client support will be 2013 and later. Office 2013 may be out of mainstream support by the time Exchange 2019 RTMs, but while it's acceptable to require the latest operating system for a new Exchange version, it's an entirely different beast to require customers to upgrade the entire Office client deployment. I anticipate Office 2019 will not be released early enough for Microsoft to require N-1 clients for Exchange 2019, and so they'll retain support for Outlook 2013 to appease customers, perhaps with an update required for best compatibility and performance.
Outlook on the web will be updated to the latest Exchange Online user experience (they're pretty close today but the EXO version will continue to evolve between now and then).
Microsoft will squeeze some more performance improvements out of the database engine. Exchange 2016 improved storage IOPS and search performance, and recently increased the maximum hardware spec that you can scale a single Exchange server up to. Given the scale that Exchange Online runs at, Microsoft is no doubt always working on ways to improve these even further, and on-premises customers can expect to benefit as well.
The server role architecture will remain unchanged. The database availability group (DAG) has been the high availability and site resilience model since Exchange 2010, and is the basis of HA/SR in Exchange Online. Edge Transport will also remain as an available role to meet the needs of customers.
Scalability for public folders will improve (although the limits are pretty good these days), but feature-wise they will remain largely the same.
The cross-premises challenges of hybrid deployments (e.g. calendar sharing, mailbox permissions, delegates, etc) will continue to improve.
Migrations will be seamless, as we saw with the side-by-side capability of Exchange 2013 and 2016.
What I don't expect to see are the cross-service collaboration features of Office 365 brought to on-premises customers. It's nice to imagine that Office 365 Groups will become an on-premises feature, but I think it's highly unlikely. Groups rely on cloud-only services (e.g. Azure AD, Planner, Teams), and I would assume that the engineering cost to make them work in some kind of hybrid fashion with on-premises Exchange mailboxes would simply be too much to pursue. Even if it were possible, having Groups on-premises and not benefiting from the continual development and improvements being made in the cloud would turn them into a poor cousin anyway. Better to leave Groups in the cloud where Microsoft controls the entire ecosystem and can iterate faster.
The Exchange Server 2019 news shows us that Microsoft is still committed to serving the sizeable on-premises customer base by continuing to develop and ship an on-prem server product as they continue to drive innovation in the cloud.