菜单

Announcing NVMe/TCP for ONTAP

目录

分享该页面

Ricky Martin
Ricky Martin

It’s no secret that NetApp made a major pivot to the cloud when it announced its data fabric vision. Our leadership and success in helping our customers make their transition to cloud is clear from the triple-digit growth rate for our cloud business and because NetApp® ONTAP® software is available in every major public cloud. Furthermore, both Microsoft Azure and Google Cloud provide NetApp technology directly as part of their native consoles. It’s also no secret that those same cloud providers don’t provide Fibre Channel (FC) as part of their service offerings, even if it is awesome. 

It’s true that there’s still steady demand for traditional FC infrastructure, and a lot of that infrastructure will be transitioning to NVMe over Fibre Channel (NVMe/FC). But most virtualization and cloud folk see it as a legacy technology, compared to Ethernet, TCP, and the rest of the internet protocol family that is the foundation of cloud. This legacy view is why many architects who want to bring the best of cloud to their data centers don’t usually include FC in their software-defined private cloud plans. That’s especially true when those private clouds are being built on top of hyperconverged infrastructure (HCI). Although HCI isn’t new, private cloud architects are becoming more aware of the inefficiencies inherent in a purist approach to it for storage-heavy workloads like transaction processing, analytics, and machine learning. This awareness has led to a rise in disaggregated HCI or asymmetric storage configurations. If neither of these terms is familiar, in essence they mean adding dedicated storage devices to an HCI server farm, an approach that should be very familiar to most dedicated storage teams. 

Although NVMe over Fabrics (NVMe-oF) has clear benefits for disaggregated HCI configurations, that fact isn’t enough to stimulate an appetite for new FC. Adding dedicated storage is a step too far for many, so asking the storage team to implement a new dedicated SAN simply isn’t going to happen. For most, iSCSI has been the go-to choice to solve this problem, but if you’ve read the first and second blogposts in this series, you’ll see that the era of SCSI is rapidly coming to an end. RDMA over Converged Ethernet (RoCE) might be a reasonable alternative, but unfortunately, as with FCoE, many of the networks underpinning these HCI-based private clouds aren't configured for RoCE. It might be possible to make RoCE work, but often the configuration changes aren’t trivialhttps://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.storage.doc/GUID-B764140D-BCF3-4C99-8169-E5B058757518.html , and many customers simply don’t want to mess with the network for a mission-critical private cloud.

This is where NVMe/TCP comes in

We don’t often preannounce technology, but this one is too good to keep under our hat. In a future release of ONTAP, NetApp plans to include a fully supported NVMe over TCP (NVMe/TCP) target. This support allows almost all the benefits of NVMe/FC, while radically simplifying the networking requirements. No need to deploy FC; no need to triple-check your data center bridging (DCB) and flow-control setting or network card firmware across your network or server farm. If you can run TCP, you can run NVMe/TCP. To take a deep dive, check out What NVMe/TCP Means for Networked Storage. If you just want the CliffsNotes version, then the bullet points on slide 16 cover it nicely. According to that slide, TCP is:

  • Ubiquitous—it runs on everything everywhere
  • Well understood—TCP is probably the most common transport
  • High performance—it delivers excellent performance scalability
  • Well suited for large-scale deployments and longer distances
  • Actively developed—maintenance and enhancements are developed by major players
  • Able to inherently support in-transit encryption

It’s like iSCSI in some ways, but better in almost every respect. “With faster-than-expected adoption of NVMe-based all-flash arrays in recent years, new technologies like NVMe over Fabrics (NVMe-oF) will continue to fuel the evolution of the enterprise storage industry. NVMe/TCP is expected to be a key technology to drive mainstream market adoption due to its ubiquity and ease of deployment. Because it is based on Ethernet, it doesn't require new hardware investment. It is particularly attractive for hybrid-cloud deployments.” said Eric Burgener, Research Vice President, Infrastructure Systems, Platforms and Technologies Group at IDC.

 

The support for NVMe/TCP in ONTAP once again showcases NetApp’s vision to create new technologies that can be deployed across the hybrid cloud. What else would you expect from a cloud-led, data-centric software company? The great thing is that this support isn’t just a quick fix bolted onto the side; it’s going to be a built-in part of ONTAP. This means customers will gain a whole stack of benefits from all the cloud and data center functionality that comes with the NetApp ecosystem. The following diagram illustrates NVMe hybrid cloud use cases.


 

These benefits include:

  • Protect against soft failure, user error, and ransomware with application-aware NetApp Snapshot copies and clones.Reduce risk via interoperability with the most commonly deployed applications, operating system, virtualization, and cloud offerings.
  • Ease disaster recovery and data mobility with the NetApp SnapMirror® feature to provide synchronous and asynchronous replication between data centers and across clouds.
  • Accelerate test and development lifecycles for mission-critical applications and databases. A truly unified infrastructure allows replication of traditional bare-metal database configurations running NVMe/FC namespaces at one location to a hybrid cloud configuration running NVMe/TCP at another.
  • Guarantee service-level availability with proven reliability and superior quality-of-service features.
  • Get peace of mind thanks to the hardened, multitenant, rock-solid security that underpins the ONTAP Security Technical Implementation Guide for the U.S. Department of Defense. 

Where is the rest of the industry?

I’m expecting that Pure will feel the pressure to say “me too” soon, though given the lackluster performance of their existing NVMe-oF solutions which I outlined in the previous blog in this series, I wonder if many customers will notice much difference.

As far as Dell is concerned, they finally came to the NVMe/FC party for PowerMax in late 2019 https://www.theregister.com/2019/09/10/fcnvme_optane_dell_powermax/ along with a bunch of caveatshttps://www.delltechnologies.com/asset/en-ie/products/networking/industry-market/h18009-fc-nvme-deployment-considerations-and-best-practices-wp.pdf, and it took them a year to push out NVMe/FC along with filling in other missing functionality into PowerStore, and there is no sign of NVMe/TCPhttps://www.dell.com/support/kbdoc/en-au/000130110/powerstore-info-hub-product-documentation-videos#Zero. Outside of that, Dell still hasn’t delivered any NVMe functionality for Unity two years after they talked about it being NVMe ready. Based on that it looks like anyone who bought a Dell Unity array may need a forklift upgrade before they get the benefits of end-to-end NVMe.

There are some other startups already talking up NVMe/TCP, in the same way that Pure talked about the benefits of DirectFlash. But none of them seem close to shipping what we’d consider to be a cloud-enabled, enterprise-ready product with all the built-in features that make it easy and secure enough to use for large-scale private and hybrid clouds.

So, where to next?

NetApp is extending our leadership in the race to end-to-end NVMe, and we’re doing that by providing an easy upgrade path to NVMe/TCP in a future release of ONTAP. This software upgrade preserves the value of our customers’ existing investments while supporting all the built-in data protection and automation features that make ONTAP easy and safe to integrate into your private and hybrid cloud plans. We want to help you deploy secure, future-proof on-premises investments for a cloud-next world, because wherever you are today isn’t necessarily where you’ll be tomorrow. 

If you want to know more, head over to the NetApp NVMe benefits and use cases page, or check out why NetApp is best for flash. But if you really want to get going quickly, reach out to one of our experts who will be happy to help you get the best out of your existing infrastructure.

Legal Disclaimer This document and NetApp's strategy and possible future developments, products and or platforms directions and functionality are all subject to change without notice. The information in this document is not a commitment, promise or legal obligation to deliver any material, code or functionality. This document is provided without a warranty of any kind, either express or implied, including but not limited to, the implied warranties of merchantability, fitness for a particular purpose, or non-infringement. This document is for informational purposes and may not be incorporated into a contract. NetApp assumes no responsibility for errors or omissions in this document. NetApp has no obligation to pursue any course of business outlined in this document or any related presentation, or to develop or release any functionality mentioned herein.

Ricky Martin

Ricky Martin leads NetApp’s global market strategy for its portfolio of hybrid cloud solutions, providing technology insights and market intelligence to trends that impact NetApp and its customers. With nearly 40 years of IT industry experience, Ricky joined NetApp as a systems engineer in 2006, and has served in various leadership roles in the NetApp APAC region including developing and advocating NetApp’s solutions for artificial intelligence, machine learning and large-scale data lakes.

查看 Ricky Martin 发布的所有文章

更多资源

Drift chat loading