Preserving data integrity in case of unplanned power events

The challenge

Over the last few years, VxRail has evolved significantly -- becoming an ideal platform for most use cases and applications, spanning the core data center, edge locations, and the cloud. With its simplicity, scalability, and flexibility, it’s a great foundation for customers’ digital transformation initiatives, as well as high value and more demanding workloads, such as SAP HANA.

Running more business-critical workloads requires following best practices regarding data protection and availability. Dell Technologies specializes in data protection solutions and offers a portfolio of products that can fulfill even the most demanding RPO/RTO requirements from our customers. However, we are probably not giving enough attention to the other area related to this topic: protection against power disturbances and outages. Uninterruptible Power Supply (UPS) systems are at the heart of a data center’s electrical systems, and because VxRail is running critical workloads, it is a best practice to leverage a UPS to protect them and to ensure data integrity in case of unplanned power events. I want to highlight a solution from one of our partners – Eaton.

The solution

Eaton is an Advantage member of the Dell EMC Technology Connect Partner Program and the first UPS vendor who integrated their solution with VxRail. Eaton’s solution is a great example of how Dell Technologies partners can leverage VxRail APIs to provide additional value for our joint customers. Having integrated  Eaton’s Intelligent Power Manager (IPM) software with VxRail APIs, and leveraged Eaton’s Gigabit Network Card, the solution can run on the same protected VxRail cluster. This removes the need for additional external compute infrastructure to host the power management software - just a compatible Eaton UPS is required.

The solution consists of:

• VxRail version min. 4.5.300, 4.7.x, 7.0.x and above
• Eaton IPM SW v 1.67.243 or above
• Eaton UPS – 5P, 5PX, 9PX, 9SX, 93PM, 93E, 9PXM
• Eaton M2 Network Card FW v 1.7.5
• IPM Gold License Perpetual

The main benefits are:

• Preserving data integrity and business continuity by enabling automated and graceful shutdown of VxRail clusters that are experiencing unplanned extended power events
• Reducing the need for onsite IT staff with simple set-up and remote management of power infrastructure using familiar VMware tools
• Safeguarding the VxRail system from power anomalies and environmental threats

How does it work?

It’s quite simple (see the figure below). What’s interesting and unique is that the IPM software, which is running on the cluster, delegates the final shutdown of the system VMs and cluster to the card in the UPS device, and the card uses VxRail APIs to execute the cluster shutdown.

Figure 1. Eaton UPS and VxRail integration explained

Summary

Protection against unplanned power events should be a part of a business continuity strategy for all customers who run their critical workloads on VxRail. This ensures data integrity by enabling automated and graceful shutdown of VxRail cluster(s). Eaton’s solution is a great example of providing such protection and how Dell Technologies partners can leverage VxRail APIs to provide additional value for our joint customers.

Additional resources:

Eaton website: Eaton ensures connectivity and protects Dell EMC VxRail from power disturbances

Brochure: Eaton delivers advanced power management for Dell EMC VxRail systems

Blog post: Take VxRail automation to the next level by leveraging APIs

Author: Karol Boguniewicz, Senior Principal Engineer, VxRail Technical Marketing

Twitter: @cl0udguide 

HCI networking made easy (again!). Now even more powerful with multi-rack support.

The Challenge

Network infrastructure is a critical component of HCI. In contrast to legacy 3-tier architectures, which typically have a dedicated storage and storage network, HCI architecture is more integrated and simplified. Its design allows you to share the same network infrastructure used for workload-related traffic and inter-cluster communication with the software-defined storage. Reliability and the proper setup of this network infrastructure not only determines the accessibility of the running workloads (from the external network), it also determines the performance and availability of the storage, and as a result, the whole HCI system.

Unfortunately, in most cases, setting up this critical component properly is complex and error-prone. Why? Because of the disconnect between the responsible teams. Typically configuring a physical network requires expert network knowledge which is quite rare among HCI admins. The reverse is also true: network admins typically have a limited knowledge of HCI systems, because this is not their area of expertise and responsibility.

The situation gets even more challenging when you think about increasingly complex deployments, when you go beyond just a pair of ToR switches and beyond a single-rack system. This scenario is becoming more common, as HCI is becoming a mainstream architecture within the data center, thanks to its maturity, simplicity, and being recognized as a perfect infrastructure foundation for the digital transformation and VDI/End User Computing (EUC) initiatives. You need much more computing power and storage capacity to handle increased workload requirements.

At the same time, with the broader adoption of HCI, customers are looking for ways to connect their existing infrastructure to the same fabric, in order to simplify the migration process to the new architecture or to leverage dedicated external NAS systems, such as Isilon, to store files and application or user data.

A brief history of SmartFabric Services for VxRail

Here at Dell Technologies we recognize these challenges. That’s why we introduced SmartFabric Services (SFS) for VxRail. SFS for VxRail is built into Dell EMC Networking SmartFabric OS10 Enterprise Edition software that is built into the Dell EMC PowerSwitch networking switches portfolio. We announced the first version of SFS for VxRail at VMworld 2018. With this functionality, customers can quickly and easily deploy and automate data center fabrics for VxRail, while at the same time reduce risk of misconfiguration.

Since that time, Dell has expanded the capabilities of SFS for VxRail. The initial release of SFS for VxRail allowed VxRail to fully configure the switch fabric to support the VxRail cluster (as part of the VxRail 4.7.0 release back in Dec 2018). The following release included automated discovery of nodes added to a VxRail cluster (as part of VxRail 4.7.100 in Jan 2019).

The new solution

This week we are excited to introduce a major new release of SFS for VxRail as a part of Dell EMC SmartFabric OS 10.5.0.5 and VxRail 4.7.410.

So, what are the main enhancements?

• Automation at scale
   Customers can easily scale their VxRail deployments, starting with a single rack with two ToR leaf switches, and expand to multi-rack, multi-cluster VxRail deployments with up to 20 switches in a leaf-spine network architecture at a single site. SFS now automates over 99% (!) of the network configuration steps^* for leaf and spine fabrics across multiple racks, significantly simplifying complex multi-rack deployments.
• Improved usability
   An updated version of the OpenManage Network Integration (OMNI) plugin provides a single pane for “day 2” fabric management and operations through vCenter (the main management interface used by VxRail and vSphere admins), and a new embedded SFS UI simplifying “day 1” setup of the fabric.
• Greater expandability
   Customers can now connect non-VxRail devices, such as additional PowerEdge servers or NAS systems, to the same fabric. The onboarding can be performed as a “day 2” operation from the OMNI plugin. In this way, customers can reduce the cost of additional switching infrastructure when building more sophisticated solutions with VxRail.

Figure 1. Comparison of a multi-rack VxRail deployment, without and with SFS

Solution components

In order to take advantage of this solution, you need the following components:

• At a minimum a pair of supported Dell EMC PowerSwitch Data Center Switches. For an up-to-date list of supported hardware and software components, please consult the latest VxRail Support Matrix. At the time of writing this post, the following models are supported: S4100 (10GbE) and S5200 (25GbE) series for the leaf and Z9200 series or S5232 for the spine layer. To learn more about the Dell EMC PowerSwitch product portfolio, please visit the PowerSwitch website.
• Dell EMC Networking SmartFabric OS10 Enterprise Edition (version 10.5.0.5 or later). This operating system is available for the Dell EMC PowerSwitch Data Center Switches, and implements SFS functionality. To learn more, please visit the OS10 website.
• A VxRail cluster consisting of 10GbE or 25GbE nodes, with software version 4.7.410 or later.
• OpenManage Network Integration (OMNI) for VMware vCenter version 1.2.30 or later.

How does the multi-rack feature work?

The multi-rack feature is done through the use of the Hardware VTEP functionality in Dell EMC PowerSwitches and the automated creation of a VxLAN tunnel network across the switch fabric in multiple racks.

VxLAN (Virtual Extensible Local Area Network) is an overlay technology that allows you to extend a Layer 2 “overlay” network over a Layer 3 (L3) “underlay” network by adding a VxLAN header to the original Ethernet frame and encapsulating it. This encapsulation occurs by adding a VxLAN header to the original Layer 2 (L2) Ethernet frame, and placing it into an IP/UDP packet to be transported across the L3 underlay network.

By default, all VxRail networks are configured as L2. With the configuration of this VxLAN tunnel, the L2 network is “stretched” across multiple racks with VxRail nodes. This allows for the scalability of L3 networks with the VM mobility benefits of an L2 network. For example, the nodes in a VxRail cluster can reside on any rack within the SmartFabric network, and VMs can be migrated within the same VxRail cluster to any other node without manual network configuration.

Figure 2. Overview of the VLAN and VxLAN VxRail traffic with SFS for multi-rack VxRail

This new functionality is enabled by the new L3 Fabric personality, available as of OS 10.5.0.5, that automates configuration of a leaf-spine fabric in a single-rack or multi-rack fabric and supports both L2 and L3 upstream connectivity. What is this fabric personality? SFS personality is a setting that enables the functionality and supported configuration of the switch fabric.

To see how simple it is to configure the fabric and to deploy a VxRail multi-rack cluster with SFS, please see the following demo: Dell EMC Networking SFS Deployment with VxRail - L3 Uplinks.

Single pane for management and “day 2” operations

SFS not only automates the initial deployment (“day 1” fabric setup), but greatly simplifies the ongoing management and operations on the fabric. This is done in a familiar interface for VxRail / vSphere admins – vCenter, through the OMNI plugin, distributed as a virtual appliance.

It’s powerful! From this “VMware admin-friendly” interface you can:

• Add a SmartFabric instance to be managed (OMNI supports multiple fabrics to be managed from the same vCenter / OMNI plugin).
• Get visibility into the configured fabric – domain, fabric nodes, rack, switches, and so on.
• Visualize the fabric and the configured connections between the fabric elements with a “live” diagram that allows “drill-down” to get more specific information (Figure 3).
• Manage breakout ports and jump ports, as well as on-board additional servers or non-VxRail devices.
• Configure L2 or L3 fabric uplinks, allowing more flexibility and support of multiple fabric topologies.
• Create, edit, and delete VxLAN and VLAN-based networks, to customize the network setup for specific business needs.
• Create a host-centric network inventory that provides a clear mapping between configured virtual and physical components (interfaces, switches, networks, and VMs). For instance, you can inspect virtual and physical network configuration from the same host monitoring view in vCenter (Figure 4). This is extremely useful for troubleshooting potential network connectivity issues.
• Upgrade SmartFabric OS on the physical switches in the fabric and replace a switch that simplifies the lifecycle management of the fabric.

Figure 3. Sample view from the OMNI vCenter plugin showing a fabric topology

To see how simple it is to deploy the OMNI plugin and to get familiar with some of the options available from its interface, please see the following demo: Dell EMC OpenManage Network Integration for VMware vCenter.

OMNI also monitors the VMware virtual networks for changes (such as to portgroups in vSS and vDS VMware virtual switches) and as necessary, reconfigures the underlying physical fabric.

Figure 4. OMNI – monitor virtual and physical network configuration from a single view

Thanks to OMNI, managing the physical network for VxRail becomes much simpler, less error-prone, and can be done by the VxRail admin directly from a familiar management interface, without having to log into the console of the physical switches that are part of the fabric.

Supported topologies

This new SFS release is very flexible and supports multiple fabric topologies. Due to the limited size of this post, I will only list them by name:

• Single-Rack – just a pair of leaf switches in a single rack, supports both L2 and L3 upstream connectivity / uplinks – the equivalent of the previous SFS functionality
• (New) Single-Rack to Multi-Rack – starts with a pair of switches, expands to multi-rack by adding spine switches and additional racks with leaf switches
• (New) Multi-Rack with Leaf Border – adds upstream connectivity via the pair of leaf switches; this supports both L2 or L3 uplinks
• (New) Multi-Rack with Spine Border - adds upstream connectivity via the pair of leaf spine; this supports L3 uplinks
• (New) Multi-Rack with Dedicated Leaf Border - adds upstream connectivity via the dedicated pair of border switches above the spine layer; this supports L3 uplinks

For detailed information on these topologies, please consult Dell EMC VxRail with SmartFabric Network Services Planning and Preparation Guide.

Note, that SFS for VxRail does not currently support NSX-T and VCF on VxRail.

Final thoughts

This latest version of SmartFabric Services for VxRail takes HCI network automation to the next level and solves now much bigger network complexity problem in a multi-rack environment, compared to much simpler, single-rack, dual switch configuration. With SFS, customers can:

• Reduce the CAPEX and OPEX related to HCI network infrastructure, thanks to automation (reducing over 99% of required configuration steps^* when setting up a multi-rack fabric), and a reduced infrastructure footprint
• Accelerate the deployment of essential IT infrastructure for their business initiatives
• Reduce the risk related to the error-prone configuration of complex multi-rack, multi-cluster HCI deployments
• Increase the availability and performance of hosted applications
• Use a familiar management console (vSphere Client / vCenter) to drive additional automation of day 2 operations
• Rapidly perform any necessary changes to the physical network, in an automated way, without requiring highly-skilled network personnel

Additional resources:

VxRail Support Matrix

Dell EMC VxRail with SmartFabric Network Services Planning and Preparation Guide

Dell EMC Networking SmartFabric Services Deployment with VxRail

SmartFabric Services for OpenManage Network Integration User Guide Release 1.2

Demo: Dell EMC OpenManage Network Integration for VMware vCenter

Demo: Expand SmartFabric and VxRail to Multi-Rack

Demo: Dell EMC Networking SFS Deployment with VxRail - L2 Uplinks

Demo: Dell EMC Networking SFS Deployment with VxRail - L3 Uplinks

Author: Karol Boguniewicz, Senior Principal Engineer, VxRail Technical Marketing

Twitter: @cl0udguide 

*Disclaimer: Based on internal analysis comparing SmartFabric to manual network configuration, Oct 2019.  Actual results will vary.