vSphere 6: “Brave New IT”

This week VMware introduced plenty of new, long-awaited product versions during an online event. I also joined this event to see the further development that was taking place since VMworld 2014.

vSphere 6

The biggest highlight was vSphere 6 – in accordance with VMware this is the “biggest release ever“. With this, a new major release is introduced after tight 4 years. 2 years ago the latest minor release 5.5 was released.

The changelog covers 650 new features – I will explain some of them in this post.

With a new VMHW (Virtual Machine Hardware Version) 11 some limitations were overhauled:

vSphere 6 vSphere 5.5
VMHW version 11 10
Maximum cluster nodes 64 32
Maximum VMs in a cluster 8000 4000
Maximum logical CPUs per node 480 320
Maximum memory per node 12 TB 4 TB
Maximum VMs per node 1000 512
Maximum vCPUs per node 128 64
Maximum RAM per VM 4 TB 1 TB

 

These changes are making SAP HANA more interesting as it is now possible to implement bigger systems. Currently SAP certifies vSphere 6 as reference platform.

Beyond the updated support of recent guest operating systems the amount of supported virtual serial ports has been increased to 32. This might be interesting for special industrial applications. vNUMA has been overhauled for better scaling in combination with “Hot Add Memory“.

Instant Clone is a new function which especially might improve VDI (Virtual Desktop Infrastructure) and automated development environments. This technique offers clones within seconds. This is implemented by creating a new VM which shares the ressources with the formerly VM in read mode. Differences are stored in delta ressources – this uses Copy-On-Write technology.

vCenter Server Appliance

I really appreciate the changes of the vCSA. For the first time this appliance is in no way inferior to the classical Windows Server counterpart giving the customer the free choice of both architectures. This can be seen in the limitations – for the first time there are no serious disadvantages anymore. Previously it was only possible to implement smaller system landscapes with the Linux-based appliance which made it quite uninteresting for big customers. The only disadvantage: vUM (VMware vSphere Update Manager) is still only available as Windows application. This means that you still need to install a separate Windows server and buy an adequate license when using the vCSA.

vCSA 6 vCSA 5.5
Maximum ESXi josts 1000 100
Maximum VMs 10000 3000
Linked Mode ja nein
vCenter Heartbeat EOL (End of life)
Linux distro SLES 11 SP3 SLES 11 SP2/3

 

vSphere Web Client

Another big update was applied to the disputed vSphere Web Client. Previsons versions were criticized concerning the speed and usability by several customers. Migrating from the conventional vSphere Client was not easy due to the new client workflow. In numerous satisfaction surveys VMware was focussing the criticism and announced the following improvements in comparsion with the previous version:

  • 100 particular usability improvements
  • 5x faster UI responses
  • 13x fast login processes
  • 4x faster context menus
  • particular actions at least 50% faster
  • consistent submenus independent of selected context

In accordance with VMware this is the first time vSphere Web Client is comparable with the conventional vSphere client. To make migrations easier the task list was relocated in the bottom of the tool. The particular parts of the user-interface are movable giving the administrator more personalization options.

vMotion

vMotion was heavily overhauled in vSphere 6. For the first time online migrations between independent vCenter servers are possible. Shared storage is not required for this and also the vCenter architecture is not relevant. So it is possible to migrate a VM from a Windows vCenter to vCSA (and vice versa).

During a migration settings like MAC addresses and DRS/HA configurations are retained. Known MAC addresses are blacklisted in vCenter to avoid duplicates. Source and target hosts need to have OSI-Layer 2 connectivity.

There is also a new interesting feature called “Long-distance vMotion” which enables migrations over network connections with higher latencies up to 100ms RTT (Round Trip Time). Using this it is possible to run clusters over higher distances (metro-cluster). This might be interesting for desaster tests and multisite load-balancing.

Fault Tolerance

When serving VMs high-available and fault-tolerant without any third-party software is requested FT (Fault Tolerance) is a technlogy chosed often. The benefit of this feature can be seen during a fault of the ESXi host serving VMs. While VMs are restarted on alternative nodes when using HA no VM fault occurs when using FT. Until now Fault Tolerance had a significant disadvantage: only one virtual CPU was supported.

With vSphere 6 Fault Tolerance supports up to 4 vCPUs and 64 GB memory for the first time. The synchronization mechanism between the primary and secondary VM was optimized. Also all three storage provisioning methods (Thick eager-zeroed, Thick lazy-zeroed and Thin-provisioned) are supported for the first time – previously you had to go for eager-zeroed Thick-Provisioning unterstützt. Using snapshots enables backing up FT-protected VMs using the vSphere Storage API. Customers can choose between vSphere Data Protection or comparable products.

High Availability

High Availability now supports VMCP (Virtual Machine Component Protection) enabling APD (All Paths Down) and PDL (Permanent Device Loss) storage faults to be recognized better. This applies to block devices (FCoE, FC, iSCSI) as well as NFS shares. If a storage fault is detected the virtual machine is restarted. Previously it was possible the storage faults did not result in a restart of the affected VM.

Beyond this High Availability now also supports:

  • up to 64 hosts und 6000 VMs
  • Virtual Volumes (VVOLs)
  • NSX
  • Network IO-Control
  • IPv6

vSphere Data Protection

Previously the vendor’s backup software was available in two editions – now there is only one version anymore. The features of the advanced edition were included in the conventional appliance. This appliance now offers up to 8 TB deduplicated storage and also offers agents for some Microsoft products:

  • Microsoft SQL Server (also clusters)
  • Microsoft Exchange
  • Microsoft Sharepoint

During restore tasks it is possible to select particular SQL databases or Exchange mailboxes.

Using Offsite Storage it is possible to save backup also in other sites. Backup integrity can be verified automatically. There are processes which restore particular VMs, starts them, checks the VMware Tools functionality and deletes the VM afterwards.

EMC DataDomain (including the function DD Boost) is supported as storage target. Using external proxies the ext4 file system and Red Hat Enterprise Linux are supported.

Multisite Content Library

Customers using multiple vCenters in different sites have been missing a utility for maintaining ISO images and VM templates in a central way. Previously it was sufficient to synchronize these files using NFS shares or other mechanisms.

This changes with Multisite Content Libraries. Templates, scripts and ISO images can be maintained in catalogs. A main catalog including those files are replicated to other vCenter cloning these changes to their local catalog. This also includes a version control system – older content is automatically deleted. Synchronization timeframes and and bandwiths can be scheduled to save valuable bandwith.

Flexible Lockdown

The Lockdown mode is used for limiting accessing the console. Once this mode is enabled the ESXi host console (DCUI, Direct Console User Interface) cannot be used anymore. vSphere now offers two modes:

  • Normal lock-down mode – DCUI is not stopped, only users mentioned in the “DCUI.Access” list are eligible
  • Strict lock-down mode – DCUI is stopped, no login possible

Using Smartcards for authentification is also new in the recent version. This feature needs conntection to  Active Directory. New esxcli commands are available for central management of local users on multiple hosts using the vCenters. For the first time it is possible to set password complexity rules for local accounts. While this needed to be implemented manually in the /etc/pam.d/passwd file previously there is now a menu “Host Advanced System Settings“.

VMware Integrated OpenStack

Using VMware Integrated Openstack it is possible to combine a pre-existing vSphere infrastructure with OpenStack to serve IaaS services. Unlike conventional OpenStack installations VMware offers a pre-configured appliance which automatically installs the needed servers – the first installation hurdle omits.

The product supports the full vSphere portfolio including:

  • vSAN
  • NSX
  • Storage Policies
  • VVOLs
  • vRealize Operations Manager
  • vRealize Log Insight

Customers of VMware vSphere (with Operations Management) and vCloud Suite are receiving the product at no extra costs. Production support is avaible for 200 USD per CPU – a minimum of 50 licenses needs to be bought. The VMware support covers the vSphere components as well as OpenStack. Additional product information can be found here: [badge style=”blue”]click me![/badge]

VSAN

Using VSAN (Virtual SAN) it is possible to use local storage of particular ESXi nodes as mirrored cluster storage (SDS, Software Defined Shared-Storage). VSAN supports two modes:

  • Hybrid – using SSD storage as Cache, hard drives as datastores
  • All-Flash – using SSD storage as cache as well as datastores

VSAN is implemented in VMkernel minimizing the overhead. Especially for hyperconverged architectures VSAN is a very interesting alternative to conventional SAN storage which is often much more expensive in comparable performance configurations.

VSAN was overhauled in vSphere 6 wurde VSAN as you can see in the limitations:

VSAN 6 All-Flash
VSAN 5.5
Hosts per cluster 64 32
VMs per host 200 100
IOPS per host 90.000 20.000
Snapshots per VM 32 2
vDisk size 62 TB 2 TB

 

There is a new feature called “rack-awareness” which places storage replicas more intelligent. Particular hosts are grouped in “fault domains” to avoid data loss forced by faulted racked. Using Storage Policies tolerated faults are defined – using this setting replicas are stored outside the rack.

A monitoring for VSAN-relevant components has been implemented – amongst others this software checks the network status, configuration, hosts and storage status.

There is also a cool fling called “Snapshot Manager“. Flings are experimental tools that have not found their way to the VMware portfolio yet – they are not supported by the VMware support. Snapshot Manager focusses on the automated creating and deletion of snapshots based on defined rules. These profiles include “Retention” rules that are well-known from backup applications. Using this it is possible to keep snapshots for a certain timeframe.

VVOL

Beyond VSAN there is another product for software-defined storageVVOL (Virtual Volumes). While VSAN is focussed on local storage media VVOL applies to SAN/NAS storage. The aim is to minimize maintenance effort of shared storage. Until now it was necessary to also execute certain administration tasks (create/mirror/assign volumes,…) on the storage system when using SAN storage this omits when using VVOL. Using the VASA API (vSphere API for Storage Awareness) native storage array functions are enabled to vSphere. This shifts creating volumes and snapshots in the storage system. When using VVOL it is also advisable to implement Storage Policies.

Conclusion

vSphere 6 introduces a lot of interesting new features. Especially for the overhauled vMotion and Fault Tolerance I can predict multiple use-cases. I’m very interested in having a deeper look at the vCSA and VADP appliances. In the first quarter of this year VMware will release the installation media for – I’m excited for first tests! 🙂

A full list including more changes can be found in the following document: [badge style=”blue”]click me![/badge]

Sharing is caring

Leave a Reply