6th Generation, what's new? So the first thing that's new is our new 3rd Generation X-Dragon Architecture. This is the architecture we use on our physical machines in order to enable our VMs to run more efficiently. At the heart of the X-Dragon Architecture is our Dragonfly Hypervisor, which is coupled with our X-Dragon chip, so we have a physical hardware device installed on our physical servers that actually serves to accelerate memory and network access for the virtual machines that are running on those physical servers.

So each physical server has this X-Dragon Subsystems that includes a chip-accelerated I/O engine, which allows for higher performance and lower latency as well as reduced network jitter. We have hardware-level quality of service management, network and storage QoS for multi-level scheduling. We also have an enhanced convergent network and enhanced hardware queuing, all as a result of installing this X-Dragon hardware on each of our physical machines.

Now we also have what we call our new Enhanced Generation 6 instances, and this is what I really want to talk about in this course, so if you look at the Enhanced Generation 6 instances, these are a combination of the Dragonfly Hypervisor plus our 3rd Generation X-Dragon Architecture plus enhanced SSD, so all Generations 6 Enhanced instances are equipped with enhanced SSD. So what does that get you? Well, it gets you better network capabilities, four times higher packet per second throughput rate than our previous Generation ECS instances at 24 million packets per second, burst bandwidth up to 10 gigabits per second, and up to 50% lower network latency.

Same thing in storage, it's a big improvement over our previous Generation ECS. All new Generation 6 Enhanced machines are equipped with enhanced SSD so storage latency has been reduced to 100 microseconds, so between 100 and 150 microseconds, and a single instance can achieve up to 480,000 I/O operations per second. This also leads to improvements in application performance.

In benchmark tests that we have run, we have a 25% improvement in running MySQL workloads, up to a 100% improvement in running NGINX workloads, and up to a 15% improvement in running Redis workloads. We've also been able to improve our availability and stability. Our single instance availability is now 99.975%, and we base our SLA on this, so we now have an industry leading SLA higher than our competitors.

Our multi-zone, multi-instance availability is now 99.995%. In terms of stability, the Dragonfly Hypervisor provides us with super lightweight virtualization and is very well-tested and very stable. It's the same hypervisor that we use to run our own VMs during Double 11, and it also has some tweaks and tuning built-in to help deal with an achieved stable performance for multi-EBS storage, so for multiple block storage devices.

If you were to perform a head-to-head between Generation 6 and the new Generation 6 Enhanced, these are the key differences, the CPU configuration is essentially the same, network bandwidth is the same but the packet per second throughput for the network cards has been improved from 6 million packets per second to 24 million packets per second, network latency has been reduced to about 25 microseconds from 32 to 53 microseconds, so things are faster.

In terms of storage, the number of I/O operations per second, we've gone from 200,000 up to about 480,000, latency, again, has been reduced for storage from 250 microseconds to read or about 160 microseconds to write, down to 200 microseconds to read and 113 microseconds on average, to write, so read and write are both faster for the new enhanced SSD, bandwidth is the same and instance level QoS is built-in on both 6th Gen and 6th Gen Enhanced, so the majority of what we have here is speed improvements, especially for our network and block storage devices.

It's obvious from the data we've seen so far that Generation 6 Enhanced and Generation 6 are a big improvement over Alibaba Clouds, other ECS instance classes are older hardware. But how do we stack up against our competitors? So it wouldn't be fair for Alibaba directly, to do that kind of analysis, so let's take a look at what a third-party thinks. So this is data from Gartner Solution Scorecard for 2020 in the compute category. Of the solution scorecard evaluates IaaS capabilities of multiple different cloud vendors, and it includes hundreds of different scoring requirements across multiple categories.

So in the 2020 Gartner Solution Scorecard, Alibaba's ECS beats AWS and is ranked 1st worldwide. This is the first time we've achieved this. We're the only cloud provider in the world as well to achieve a perfect score in both the required and preferred category for our ECS service, so we've achieved about as good a score as we can manage. Our overall score has also greatly improved since 2019, thanks to the updates and continuous improvements we've made to the ECS service. Gartner's overall conclusion for Alibaba Cloud changed from last year's suitable for specific scenarios to suitable for general scenarios. So we're now on par or even better than most of our major competitors in the cloud computing IaaS space.