During Google Cloud, Google seized the opportunity to introduce a number of new instances that can provide workloads with more powerful chips. These include the 4th generation C and N instances, C3 bare metal machines, X4 memory-optimized bare metal instances and Z3, the first storage-optimized VM family.
In addition to all those new instance types, there was also a focus on the new Google Axion processor. However, that one is not available for now and thus there is no instance type for it yet.
4th generation C and N instances
The new 4th generation C and N instances are the successors to the popular 3rd generation that are used for many workloads. Even Techzine runs on a third-generation C instance. The 4th generation is powered by a new 5th generation Intel Xeon processor. C4 and N4 are specifically designed to handle generic workloads. Where the N4 delivers a bit less peak performance and is more capable of optimizing costs.
C4 instances
According to Google, the C4 has a 19% better price/performance ratio over other cloud providers and a 25% better price/performance ratio over the C3 VM. This makes C4 well suited for databases, caching, network appliances, Web servers, online gaming, analytics and media streaming.
With C4, the host OS can also be updated and managed with live migrations without impacting customer workloads. This provides the additional benefits needed to avoid downtime.
C4 Instances are available in 24 different sizes, up to a maximum of 192 vCPUS, 1.5TB of working memory and 500K IOPs, and network speeds of up to 200Gbps can be achieved with the C4 instances.
N4 instances
The N4 is also for generic workloads, but more focused on cost optimization. According to Google, the N4 offers 18% better price/performance ratio, but, for example, a 39% better cost/performance ratio for MySQL. These are mostly workloads without real peak performance. Think dev/test environments, virtual desktops, microservices, business intelligence, code repositories, data analytics, storage and archive and medium traffic Web servers.
C3 bare-metal machines
With C3 bare-metal machines, Google is now bringing the successful C3 instantiation to bare-metal machines. This allows customers with special workloads such as specific hypervisors to also run on these powerful chips. This allows special workloads or optima use of software that is still based on a CPU license. In such a case, as a customer, you want the maximum performance with as few cores as possible. That’s where the C3 bare-metal machines can help.
One of the Google partners that will be working on this is Nutanix. They will be deploying the C3 bare-metal machines to offer Nutanix clusters from the Google Cloud. “We are looking forward to the launch of Google Cloud’s new bare metal machines types, said Tarkan Manaer, Chief Commercial Officer of Nutanix.”
X4 memory optimized instances
With the new X4 memory optimized instances, you get an instance with a powerful processor, but more importantly, you have much more working memory at your disposal. These instances can be built up to a maximum of 1920 vCPUs and 32TB of DDR5 working memory. For these instances, SAP in particular is a good partner. “SAP Enterprise Cloud Services is exploring how it can deploy X4 and C3 metal systems for Rise with SAP. SAP is impressed with the reliability, performance and ease of management, said Lalit Patil, CTO for Rise with SAP, Enterprise Cloud Services, SAP SE.”
X4 is still in preview and available in the U.S., EMEA and APAC.
Z3, the first storage-optimized instance
Google Cloud has now also chosen to introduce an instance optimized for storage. These instances use a 4th generation Intel Xeon processor and are combined with latest generation SSDs. Z3 has the highest number of IOPS both within Google Cloud and compared to other cloud providers, Google said. It equates to 6M 100% random read and 6M write IOPS. Z3 also offers three times more throughput than previous generation SSDs and 35 percent lower latency.
Z3 is available in two versions: 88vCPUs with 704GB of DDR5 working memory and 36TB of local SSD storage or 176 vCPUs, 1.4TB of working memory and 36TB of local SSD storage. This is clearly aimed at large enterprise organizations with lots of high-speed storage needs. According to Google, this is great for horizontal and scale-out databases such as Aerospike, ScyllaDB, as well as log and search analytics such as Elastic and Splunk. It can also serve data warehouses well. Z3 is available immediately.