Top 5 most futuristic projects
from Mark’s Inside Azure Datacenter Architecture session
First Spring Ignite 2021 has come to an end, but all the highlights are still shining to some degree. Our world is evolving all the time, with those conferences always good to keep our eyes open.
One of my favorite sessions which has taken place in nearly all the ignites, and is constantly being updated, is the Inside Azure Datacenter Architecture from Mark Russinovich. In the past, I’ve attended in person and it was indeed an incredibly popular and valuable session. It is an eagle angle that shows the big picture of the latest technology from Microsoft from many important perspectives. Such as this Spring session has been broken down into a few sections, for instance, Datacenter and Intelligent Infrastructure, networking and compute, storage and data, developer experience has brought more focus on Azure Resource Manager (ARM).
We’re not going to go through all the highlights in this blog, I believe we’ve all been going through an excitement by a significant number of announcements from Microsoft Azure, it’s indeed becoming better and better. However, some highlights might not appear to shine in the first place, although with all the baby steps that have been taken over the last few months, when the time comes, it will make a huge difference in the world.
Inside Servers – Mega-Godzilla-Beast
First highlight of cloud servers is super memory-optimized servers. Over the last few years, we can see an epic growth in the G family, where G stands for Godzilla. From 2014, we’ve seen huge 6 years ago now it’s relatively small. Below is the table showing the evolution path for G-series, most of the VM in this series are for the unparalleled computational performance such as SAP HANA workloads, and it is also shining in handling heavy requirements such as large databases, data warehousing etc.
Godzilla – 2014 | Beast – 2017 | Beast V2 – 2019 | Mega-Gozilla-Beast | |
vCPUs | 16 cores | 72 cores | 224 cores | 448 cores |
RAM | 512GB | 4TB | 12TB | 24TB |
Disk | 9×800 GB SSD | 4 x 2 TB NVMe1 x 960 GB SSD | 4 x 2 TB NVMe1 x 960 GB SSD | 2 x 960 GB SSD4 x 6.4 TiB PCIe |
40Gb/s | 40Gb/s | 50Gb/s | 50Gb/s |
Azure Quantum – Gooseberry quantum controller ( project cryo-CMOS )
Joined by Microsoft research and the University of Sydney, project cryo-CMOS represents a big breakthrough for solving one of the hardest problem which is the quantum control, meaning how to keep a extremely low temperature just a few millikelvin might be even lower than space cold, and stabilize the Qubits by adopting right approach in material engineering, and the Gooseberry processor was designed for large-scale quantum computation. With Quantum dot technology it’s now it’s possible to run millions of Qubits on a small wafer at the bottom of quantum refrigerator as opposed to other quantum technology which manages the QBit with Quantum computers with the size of a conference room.
Embrace DevOps – Project Bicep
Microsoft has been continuously innovating developer experience, Bicep shows how they have taken the feedback from the community and their determination to democratize ARM with other presentations that are more friendly to the developers. Bicep is a Domain Specific Language (DSL) which allows to authoring Azure resources in a declarative fashion, technically, any features available in ARM templating authoring are available in Bicep too, and it’s transparent to each other, that being said, if you have an ARM template, it can be easily translated to bicep and it also works in the reverse direction. It will significantly change the developer experience. Instead of the only possibility in the past working with JSON-based files, they can now use declarative language to build the same infrastructure that they’re already familiar with, and it’s especially beneficial while working with complex infrastructure deployment and configurations. Nice to know there are better options out there instead of using nested ARM templates.
Intelligent Data – Confidential computing with SQL Server
Besides Data encryption at rest and in transit, the next move for data protection has been proved to encrypt data ‘in use’, that is where confidential computing comes in. The confidential computing uses a concept called ‘Enclave’, where data is going to execute with the trusted code of the application. Microsoft actually integrated this technology into SQL Server, and with the DC-series of SGX hardware with SGX enclaves. It will allow the SQL query processor to run inside of a trusted enclave, and then release secrets to it that allows it to decrypt the encrypted parts of the database and perform rich computation inside of it, and this happens on the server-side rather than previously the always encrypted happened on the client-side.
Storage innovation – Project Palix
When it comes to storage, we’re always in the pursuit of better storage technology, for ever-increasing amounts of data with more durability, for longer periods of time without having to be rewritten. In the past, Mark’s session was focused on a storage technology under the umbrella of project Silica which aims to store data in the glass. Highlights this year is in particular futuristic when he presents project Palix cooperate by Microsoft research and the University of Washington.
The Project Palix is about storing data inside of DNA molecules. You might be wondering how this could happen and why. DNS molecules have the benefit that they’re extremely durable, some DNA discovered in the fieldwork of paleontologists have existed on the planet for millions of years, that being said DNA is stored with a proper external condition, it lasts almost forever. Another great feature of DNA is because it’s super dense that it could storage a zettabyte of data inside of a rack, which is just a small facility within a standard datacenter. Therefore a ‘rack of DNA storage’ could store the same amount of data that costs hundreds of datacenters today.
Mark explains the technology is to synthesize molecules to encode bitstreams into the DNA, and then simply store the DNA under appropriate conditions. The way to read the data is also a standard DNA sequencing technologies to read those DNS molecules out and retrieve the original data.
If you want to know more details about this project, you can go check this session. There’s actually a demo where Mark showed the DNS storage which he extracted using the synthesis system, looks pretty cool !
Conclusion
Here are the top 5 most futuristic projects that I read from Mark’s Inside Azure Datacenter Architecture session from this year Microsoft Spring Ignite, it actually shows a huge landscape that Microsoft has been investigating and innovating in the technology, which is not only in the cloud computing area but also how those technology emerging, innovating each other, how those fundamental technology then apply them to different industries, and when the time comes, they will play a vital role in our future life and without seeing them coming.