This is a wrap-up of an Ignite session (hosted by Bill Baer) which provided a very useful insight into what we can expect with the release of SharePoint 2016…..
Cloud investment benefiting on-premises
- Traditionally, the release of an on-premises version of SharePoint has provided the definition and foundation of the cloud offerings for SharePoint Online and Office 365. But now, the roles have been reversed and MS are taking what they’ve learn from the cloud, and on a hyper-scale, into SharePoint 2016. SharePoint 2016 is a ‘point-in-time’ snapshot of SharePoint Online.
- SharePoint 2016 is the most durable, robust and tested version ever released due to its presence in the cloud. It’s a comprehensive solution for integrated and connected information.
New management investments:
Hardware (pretty much on par with SP2013 requirements)
Perquisites (Hardware & Software)
- App Fabric (varying caching capabilities) still a ‘supported’ component of SharePoint despite going out of Windows support next year
- SQL 2014 minimum database requirement
- No support for SQL Express and standalone SharePoint installations
Upgrade & Migration
- New version (16.0.4021.1201)
- Only upgrade from SP2013 to SP2016 supported – no direct upgrade from 2010. The number of architectural changes in SP2013 won’t allow this version to be bypassed (best upgrade experience)
- Any SP2013 14.5 Site Collections (effectively running in SP2010 mode, will need to be brought forward to SP2013 before they can be migrated to SP2016. Alternatively, SP2013 sites can be migrated directly to SP2016 or you can use database attach methodology.
- No significant Service Application architecture changes. Performance Point available in SP2016
- SharePoint Designer will not be updated for SP2016 –SP2013 version still supported
- With SP2013 Windows claims, Forms-Based, WSFED, SAML Claims to name but a few, were all made available to provide backwards capability. With SP2016 SAML authentication now treated as a ‘first class citizen’ and OAuth and this will open identity options which weren’t available in SP2013 when it certain scenarios (Business Intelligence)
- Apps will trust Azure Active Directory
Older identity models will be supported (Windows SAML) but just like SP2013 when ‘Classic’ became deprecated, this is the first move away from domain identity and towards cloud identity.
- With SP2010 & SP2013 all traffic using Port 25 (Alerts, Reminders) was unencrypted. Now, non-standard ports can be leveraged and encrypted using STARTTLS – encryption between SharePoint servers and messaging systems
- If encryption fails on non-standard port, it will not fail-back to an non-encrypted port – process will fail. All configurable via Central Admin and Powershell
Performance and Reliability
SP20103 roles were quite agnostic and there were four types of different roles (Distributed Cache, Request Management, Web Servers, Batch Processing). Roles were ultimately defined by guidance in Technet documentation. In SP2016, ‘roles’ have been defined by code – there are effectively three:
- User Services. This is a
Web server receiving any form of request from a user. In SP2013, this request may have to ‘traverse’ a number of layers (Web/App Servers) to meet the request and then ‘traverse’ back up the stack to the Web Server before providing the result to the client – so, potential degree of latency to each request.
- Robot Services – any request that NOT initiated by end user (Timer Jobs, Batch processing jobs)
Caching Services – distributed cache
Any SharePoint request, whether internal or by end user, should be managed by a single role server. Improve performance and efficiency and crucially, avoid the traversing of requests through multiple layers.
- ‘Specialised load’ role > third-party solution which provides a role within SP Farm > could be a non-standard Service Application. Now we can scale a server on a unit role basis whereas previously, server may had had no identity
- Single server Farm’ – still need a SQL instance, no SQL Express availability
- New ‘health rules’ introduced to maintain new ‘min role based rules – detects any deviation away from role to ensure topology compliance is maintained
- Powershell can be used to provision server roles for scripted build requirements
In the past, number of .MSI per update, multiplied when language packs required. SP2016 patching, reduced number of .MSI and .MSP to reduce footprint – so patch is smaller and optimised performance. Execution time faster and with ZERO downtime.
All of the upgrades now run ‘online’ as opposed to ‘offline’ which subsequently led to the stopping/starting of services. Now, you can effectively patch SharePoint during the middle of the day! This ability to patch online comes from SharePoint Online with agreessive SLA’s which need to be adhered.
- Improved reliability and performance
- New feature which switches off the NTLM authentication between SharePoint and the Distributed Cache cluster which previously caused failures and delays – authentication overheard as the constant chatter with Active Directory. New transport layer mitigates against these delays.
Boundaries & Limits
- New scalability by virtue of SharePoint Online thresholds
- Increased content DB size (likely to be into the TB’ as opposed to GB)
- 100,0000 site collections per content database
- List threshold increased to 5,000 items
- 10GB upload max file size
- Search increased x 2 > 500 million items in Search scale
Fast Site Creation
End users don’t generally have self-service Site creation which from a SP perspective is a heavy-lifting operation (provisioning and enabling of features/services) .Now using SPSite.Copy at Content Database Level you can rapidly enable site creation.
In the background, we’re replicating SQL rows from a ‘master’ template from point A to point B – and thus bypassing code and feature activation
User Profile Service
- No longer ‘baked’ into SP
- AD synchronisation rejuvenated (old 2007 SP way!) or use external replication via FIM installed on a separate server outside of SharePoint. No more FIM!
- Project Server now part of the conventional SP Content Databases – more of a consolidated approach
db’s into Content Database
- Project Server still needs to be licensed independently even though it is embedded into SharePoint
In early versions of SharePoint, if you sent out a link/URL to an item, and then renamed or moved the item to another location, the link become broken and unusable. This is resolved in SP2016 with resource-ID based URL’s, which work on endpoints.
New files when uploaded are provided with a ‘resource link/site ID’ and an affinity – so, regardless of whether the file is moved to, either a new Site Collection, the affinity is maintained.
Or, if the file is renamed, the underlying link doesn’t change because the file is marked with a resource ID and an affinity – it doesn’t matter where the new URL is.
- Unified compliance portal across online and on-premises
- Document fingerprinting, item-level encryption (available via hybrid connectivity – also compliant with Exchange and Skype for Business)
- Azure RMS for SP on-premises and Online (with single item encryption)
- EDiscovery hybrid offering across both online and on-premises
Cloud Search Service Application
New Service Application which will be shipped later this year as part of SP2013 (and fully embedded within SP2016). This will unify cloud and on-premises search indexes – single entity for Search. This will allow you can search from Cloud or on-premises into single results pane.
Brings the power of Office Graph/Delve to on-premises (*Delve is not deployed on-premises’ – you leverage the capabilities via the new Cloud Search Service*)
Extranet Site Publishing
- Publish internal sites to Internet (Extranet hybrid scenario – for people struggling with Federation options like ADFS)
Team Site Improvements
- Hybrid team sites > allow you to follow a site whether it’s on-premises or online. One unique panel across both environments
- Hybrid deployments were complex to implement (with Search, BCS). Scenario picker is an automation wizard for implementing hybrid experience – far cry from 15 pages of Powershell