Tech Bytes

I had a briefing with ReSTNSX, and they demoed their VMware NSX-enhancing platform to me. Have a listen in the player above for my impressions of the demo.

Features Include…

* A different UI for NSX that makes it easier to leverage the NSX API.

* The ability to import lots of objects via CSV instead of clicking until your finger breaks.

* Creation of custom workflows.

* Dashboards for NSX tenants.

For More Information

* http://restnsx.com

Virtual Instruments has visibility and testing tools for FibreChannel networks.

Analyse the network traffic to determine the performance of the storage traffic – file protocols and object protocols

It can analyse storage protocols  captured from the network and then replay these patterns against various storage equipment. This allows for testing and analysis of the vendor equipment.

Why is this important ? An inspirational cynicisist view is that storage vendors are unable to make products that work properly or as advertised so there is need to test and validate the operational capability. Or perhaps customers expect to see bugs in the product and testing will help to find them. Another use case is that product upgrades are expected to be unreliable so testing of new software is needed to ensure that the vendor got it right.

A less cynical view is hard to find. Storage products are not made to fit every use case and customers are not able to easily pick out the product that is good for them. Testing helps here.

Anyway the idea of load testing is well established, capture data from the network and software can then analyse the captured data and then create a replay. You don’t need to see the payload here. You can collect the storage protocol requests and then build a test that simulates.

I saw demonstrations of the software interface that showed a wide range of analysis and showed the visibility. The ability to conduct a test workload and then vary the storage query parameters.

There are similar products in the networking space – look at Ixia Systems and Spirent who do similar things for networking with the protocol inspection. Virtual Instruments offers a version of this that is specific to storage industry, seems to have deeper capabilities.

Takeaways:

* the larger part of the the storage market believes that they are special snowflakes with unique requirements and specific conditions that only storage specific solutions can solve.

* The value of storage is different, remember that a data loss is a permanent and potentially catastrophic event. A network outage has less impact and there is less we can do for true availability.

* For customers with critical dependencies on the storage engine, the ability to test and analyse storage traffic to validate storage products is key. And Virtual Instruments is a solution in this space.

 

Micron makes memory and storage.

* total 110B market for memory

* total semiconductor market 385B

* memory is a substantial proportion of the whole market.

Micron highlights two transitions from centralised storage to distributed storage, then placing the data at the edge of the network.

Interpretation 

* storage arrays are less important because they are smaller part of the mix

* edge storage is a new market for smaller, niche solution

Public cloud storage spent $19.6B in 2018. growing at 12% CAGR

F500 moving to a combined in/off premise storage model  — Some workloads and data will remain on premise

Takeaway: There will be a dirty winner takes all battle for enterprise IT storage.

SSD are $7B market growing 31% CAGR

Flash is 50% is Enterprise SSD market

Public cloud is driving the technology in SSD, while the Enterprise is moving away from storage array to distributed storage engines.

Enterprise it storage will use technology that is led by public cloud companies driving their requirements with manufacturer

Micron is making a wider range of products.

Takeaway: whats striking here is the change in storage technology now that legacy storage vendors are no longer involved. When most storage came through EMC, it made sense to prevent innovation so they could sell old products at higher products. Make something once, sell it many times is a good business model that inherently prevents innovation.

Now that public cloud/hyperscalers are driving technology and spending money to buy the newer storage, this means that a wider range of different products is possible.

So we will see SATA SSD, NVMe SSD exist. Plus other 3D express, 3D NAND etc where before it was difficult to bring them to market via the storage vendors.

Micron talks about improving the Linux software for flash. By improving the applications you can achieve order of magnitude improvements is flash. Working with database software to increase storage performance by 8 times. Also reducing Linux storage stack by 95% while reducing power consumption by up to 85%. I’d like to get more details how this is done to understand more about this but its

Takeaways: 

Storage industry is growing in new ways. The need for more storage is only part of the story, the rise of hyperscalers are driving new technology which drive new products.

At the same time, there are moves to optimise the storage software drivers so further improve performance and power consumption. Again, driven by hyperscalers companies because enterprise it vendors never really cared about power or improving their products.

Enterprise IT is slow to implement change. In my view, legacy storage vendors are slow to adopt new technology because they have to spend money (they would call it investment) to make new products. It makes more sense for them to sell the old products and get sales people to convince customers that the old stuff is still worth buying.

So the public cloud gets better at a faster rate than enterprise it. Because they are willing to implement new technology while Enterprise IT isn’t.

In this briefing, Netscout talked to me about how they are unifying visibility fabrics inside of an IT team, so that both network operations and security operations can monitor the network with common tools.

This is a big deal according to Netscout because secops folks tend to build out their own toolchains to get the functionality they desire, whereas on the Netscout platform, secops folks can share some tools with others in IT while still having their needs met.

Click in the player above to hear my overview of the briefing.

For More Information

* Netscout nGenius PFS5000 product page

* Netscout Unifies Visibility For Network And Security Ops (official press release)

I do a quick walk through of VMware’s official release notes for NSX 6.4. Click in the player above to listen to this short round up of what stuck out to me the most.

 

 

For More Information

VMware NSX 6.4 Release Notes

The final presentation from Networking Field Day 17 was unique: Mellanox, Ixia, and Cumulus shared the presentation slot to talk about their individual merits as well as how they work together.

Mellanox

Mellanox builds Ethernet and storage switches for a variety of use cases. The company touted its custom silicon as a differentiator, as well as the fact that you can run Mellanox using its own network OS, or load a variety of third-party NOSs (hint: Cumulus is one of them).

After a brief tour through its Ethernet product line, Mellanox made the case for a Clos/leaf-spine configuration as the ideal design for a data center network, and EVPN VXLAN as the ideal fabric for that design.

Ixia

Next up, Ixia presented on its flagship IxNetwork testing suite and how it can be used to validate the performance, scalability, and efficiency of a data center network.

The company followed with a demonstration of a test of a data center network built from Mellanox switches running the Cumulus NOS and using a VXLAN fabric.

Cumulus Networks

Last but not least, Cumulus presented on its Cumulus Linux network OS and how it works on whitebox switches. Cumulus touted the advantages of a Linux core, including the ability to manage the software using common tools such as Puppet and Chef.

For those not conversant with Linux, Cumulus also shared details about its Network Command Line Utility (NCLU), a CLI that network operators will find familiar.

The presentation also demonstrated some of the capabilities of Cumulus NetQ, a telemetry system that collects state information from NetQ agents running on Cumulus and other Linux OSs, bare metal hosts, and VMs, and then sends that state information to a database that engineers can query.

Using NetQ, engineers can validate network state, test configuration changes in a virtual lab, and diagnose problems.

Cisco Systems presented at Networking Field Day 17. Three broad themes emerged from the presentations: intent, automation, and telemetry.

The session began with a presentation on Cisco’s goals for the user experience and design of the company’s forthcoming intent-based products. While the discussion didn’t have much technical detail, it did demonstrated how the design of a UI needs to align with the goals of an intent-based system, including how intent is expressed, how it’s simulated and executed, and how the results are reported back to the user.

This session also generated some bracing conversation about the notion of “gamification” in a UI.

Automation

Cisco walked delegates through two demonstrations of security automation. The first was on using SD-Access to minimize the work of firewall management by pushing centralized configurations from DNA Center.

The second demonstration showed how to use Cisco ISE and TrustSec to automatically enforce security and policy controls on client devices.

Cisco also provided a deep dive into how it’s leveraging Linux, open source tools, and DevOps principles to make its IOS-XR routing platform better suited to Web-scale environments.

One way it’s doing this is by allowing third-party apps to run directly on the router. It’s also exposing APIs to make the platform more programmable, and enabling the use of tools such as Ansible, Puppet, and Chef.

Telemetry

Sticking with IOS-XR, Cisco provided a detailed demonstration of the telemetry data that can be pulled from the platform. Telemetry is key to improve visibility for operations and performance. Telemetry will also be a critical component of an intent-based system for modeling network state in real time and verifying the result of changes.

Field Area Networking

One presentation that didn’t fit neatly into the themes described above, but did have a high cool factor, was on Field Area Networking; that is, wireless mesh networks of industrial sensors and devices. Cisco is building wireless gear directly into industrial sensors for industries such as public utilities, shipping, and mining.

VMware presented at Networking Field Day 17. The presentation covered several topics, including an update on VeloCloud, NSX-T features, and a dive into the latest version of vSphere.

The Business Fabric

VMware began with a presentation from Pere Monclus, VP and CTO of the Network and Security business unit. He outlined VMware’s overarching vision to be the “business fabric” for the cloud age: that is, extending an NSX-based overlay from the data center to public cloud services such as AWS and Azure.

The company also intends to extend that fabric to the branch, thanks to VMware’s acquisition of SD-WAN vendor VeloCloud. There’s still a long way to go to integrate VeloCloud into NSX, but it seems clear that’s VMware’s goal.

Speaking of VeloCloud, delegates saw updates to that platform, including a new SD-WAN service for PCI compliance, the ability to segment different types of traffic at the branch, and other advances.

NSX-T

VMware also touted NSX-T for cloud-native environments, including an NSX container plug-in that integrates with Kubernetes and CloudFoundry.

The goal is to enable infrastructure teams to enforce networking and security policies within virutalized and container environments while letting the application developers do their thing.

NSX For vSphere 6.4

VMware provided a deep dive into the latest release of vSphere and its integration with NSX, including advances to its firewall capabilities. NSX could already function as an L4 stateful firewall; it’s now moved up the stack to L7, with the ability to recognize and enforce policies on approximately 50 common applications.

Audio Note:

If this podcast sounds like we recorded it in an empty hotel gym, that’s because we did. Network Field Day moves fast, so we had to take the first free room we could get.

Juniper Networks came back for a second day at Network Field Day 17. This time around the company focused on:

* An Appformix demo showing analytics and visibility – a topic that we say is critical to the future network strategy

* Contrail diverging from OpenContrail to embrace public clouds, hybrid clouds, and Kubernetes. It aims to be a single platform to unify all network orchestration (with some companies pursuing multiple controllers)

* Contrail as the configuration engine, AppFormix as the analytics and automation layer

* Adopting P4 across its fleet as a general abstraction for forwarding packets

* Open/R on its current control plane inside devices (and contrail vRouter)

Juniper outlined its idea of how network automation will evolve:

* Day 0:  Human-Driven Automation

* Day 1:  Event-Driven Automation

* Day N: Machine-Driven Automation

Juniper also demonstrated PeerBot – a Contrail application for automating BPG configuration across dozens of routers for IXP environments. PeerBot:

* Dramatically simplifies BGP configuration

* Targets a narrow market, but this makes it useful. A general-purpose solution might struggle to be useful to specific people

* Is just one of a handful of bots that Juniper has announced. The company plans to build more bots this year

Extreme Networks presented at Network Field Day 17 and focused on automation via StackStorm and Workflow Composer.

Extreme also discussed its ability to run a VM and applications on certain SLX models.

StackStorm Is Unique Because…

* A vendor is delivering this and supporting it

* A moderately competent network engineer will be able to do something useful with this

* The workflow helps you do something reliable

* Mean time to pay rise is drastically reduced

* Effort to achieve pay rise is reduced

Workflow Composer

Workflow Composer builds on Stackstorm to provide cross domain automation changes the way you operate the network.

It speeding up the operations lifecycle, reducing the time spent on deployment, troubleshooting, and implementation so you can focus on remediation. The vendor can help with these three; only you can handle the last part.

Check out the NFD videos to see Workflow Composer demos.

VM On A Switch

* Redirect a 10G interface to VM running on the switch

* Isolated from the control plane

Workflow Composer And VM Demo

* Using the VM on a switch

* Slack integration for ChatOps

* Capture packets on multiple switches

* Upload PCAP to cloushark and then link appears in the Slack