Multi-hypervisor workload ... what? Just think of it as a virtual MAGIC CARPET
Zerto converts VMs between hypervisor environments
Comment Zerto offers multi-hypervisor and cloud workload transformation and protection services – there, I’ve said it. But what does it mean?
It is doing this because of the coming of hybrid compute and storage clouds.
It all comes down to this question: What does a hypervisor replicator company do when hybridity is going to be a fact of its business existence? Does it judge it an opportunity or a threat?
Zerto appears to be seeing it as an opportunity, and is working to offer workload migration and transformation services that make it easy to view different hypervisor and cloud compute services as a single compute resource domain within which to run virtual machine workloads
The public cloud part of hybridity is far more than hyperscalers like Amazon, Azure and Google – the humungous trio at the top of the public cloud tree.
For Zerto it consists of many, many cloud service providers (CSPs), some coming from the telecoms market, and it counts 100 of these as customers. Three or four major European telcos are coming on board too, according to Gil Levonai, Zerto's VP for marketing and products.
Levonai says they see Zerto’s replication as the basis for a disaster-recovery-in the-cloud service they can sell to their customers. He claims Zerto replication is “replacing array-based replication in hundreds of customers, like B & Q, in banking, retail, and healthcare.”
It uses compression to reduce replication bandwidth needs, and is enterprise-class, with granular multi-tenancy, billing, resource allocation and infrastructure abstraction.
The firm says it has customers in 26 countries now for its replication and cloud offerings, with cloud typically used as a DR facility. It sells to enterprises larger than SMBs, and won't go after bids with less than 15 virtual machines (VMs) involved.
The team said it sees the data centre of the future being hybrid and having three components:
- On-premise server and storage array hardware for ERP and other business-critical applications
- Public CSP-provided facilities with SLAs, for things like disaster recovery (DR)
- Amazon and Google type CSPS for backups needed for long-term retention
Zerto’s belief is that customers want infrastructure choice (buy and build) rather than lock-in. They also typically need three copies of a workload. For example:
- ERP workload on high-end compute and storage
- Copy in the cloud, with SLAS, for disaster recovery
- Backup and long-term retention in a public cloud such as that of Google or AWS
What would be good for flexibility and as an anti-lock-in measure would be the ability to take an ESXi VM and run it under Hyper-V, or move it to AWS and run it there. What Zerto is doing is taking production workloads – such as SQL and SAP – and moving them between different (hypervisor and on-prem/cloud) infrastructures – and that’s what its customers are asking for, according to Levonai.
It can provide large-scale migration between infrastructures, such as moving 1,200 VMs as part of a data centre consolidation exercise, and preserving a specific boot order for the VMs.
The aim is to offer the orchestration of such moves, within a cloud fabric as it terms it. Zerto said it would add offsite backup as a replication target soon, with Amazon Glacier as the probable target.
The inhabitants of its cloud fabric are going to be VMware, Hyper-V, Amazon, Azure, and OpenStack. The sequence in which public CSPs get supported will be driven by customer’s demands.
It has devised the notion of a Virtual Protection Group, a collection of workloads and metadata about them, that can be moved from and to and infrastructure member of its cloud fabric.
Levonai said: “We don't want to create another abstraction layer. We want the workloads to run natively on each infrastructure... We’ll do that by converting a workload to run on the different infrastructures.”
It’s not about creating a hypervisor of hypervisors. The long-term retention will most probably include Amazon’s Glacier as a target but Zerto is agnostic: “All we need is file system access. We’ll natively support Amazon. But on top of that we’ll pump [workloads] into a file system.
“Our CSP customers should love it. Long-term retention is cheap storage that will cost them less… CSPs could protect Hyper-V using customers on the CSP’s VMware infrastructure.”
What about KVM support? “It’s possible. I don’t actually see real production workloads on it. If there’s a business case we’ll do it.”
Levonai said Hyper-V support is something Zerto’s customers want, in case, we suppose, VMware adds more pricing and licensing restrictions. They want the Hyper-V choice, the Amazon choice, to keep the VMware offering reasonable.
Cisco’s Intercloud concept is basically the same idea – connecting different clouds to flatten the network. The marketing bod pointed out that Cisco had seven technology partners in its Intercloud announcement – and Zerto was one of them.
Levonai said he sees Zerto becoming an infrastructure choice enabler, an enterprise IT service broker.
And you thought replication was boring? It’s becoming a magic carpet to move, translate and migrate, virtual workloads between competing virtualised infrastructures. And it doesn't hurt that it prevents your workload infrastructure provider from locking you in. ®
Sponsored: Benefits from the lessons learned in HPC