When The Cloud Jumps Into Real-life … Meet Snowcone

Working at the edge of the cloud

When The Cloud Jumps Into Real-life … Meet Snowcone

Working at the edge of the cloud

Well. This week I’ve taken the jump from a bare-metal server to AWS, and I think I’ve saved myself some money, as I’ve managed to get an efficient Web infrastructure, and which actually perform better. I now understand the costings of AWS, and actually it’s quite simple, and where it is disk space that costs much more than server time. But, it’s the flexibility of the cloud, and the ability to scale up and down as require that the Cloud comes to the fore.

If this COVID-19 period has shown us something, it is that we need to become more resilient, especially as we increasingly rely on the access to computing resources and networks.

Our society and organisations need to become more resilient, and especially on what happens in an emergency situation. What happens if the electrical power in a hospital fails? What if there’s a major explosion within a city centre? Overall we are becoming increasingly dependent on AWS and Azure clouds, but what happens when the connection to the cloud fails, would we still be able to operate? Well, for this we must move to the edge of the cloud, and see if we can jump our of the Cloud and into our physical environment. If you have lots of data and/or require lots of computation, but have limited bandwidth, how do you still run your cloud? And if you have an autonomous vehicle, what happens if the connection to the cloud glitches?

And so there are many applications that need to move computation to the edge of the Cloud, and for this Amazon have just released AWS Snowball Edge. It is smaller than the previously related Snowball, and is built into its own cloud environment:

And so it pushes cloud computing out of the Cloud and into a physical environment. In this way your computation — while still virtual EC2 instances — runs on an edge device. Overall it can store around 8TB of disk storage, has 2 CPUs, 4GB of memory, wifi/Ethernet, and runs on off a battery or AC power. It also has enough power to run EC2 instances.

You could just see it in an emergency situation or within a hospital, and providing a backup system to a large scale data infrastructure. It can connect to the cloud directly, but can also be used to build extra resilience with connections into the larger Snowball and into the Snowmobile truck:

The device is also tamper-proof and includes a Trusted Platform Module (TPM). This means that the data can be encrypted with the key stored in the TPM, and then using AWS KMS (Key Management Service) for transmitted data back to the Cloud.

Increasingly we will see computing move out of the cloud and back into local devices, and where there is almost zero latency, but where it is all still virtualised. This might include the processing in autonomous vehicles and drones, and where the delays involved in communicating with the cloud could cause issues. Along with this, we may have localised applications that require masses of data to be captured and stored, such as within video broadcasts. This allows for local processing, before the upload into the public cloud. It has also been designed to be robust against harsh environments, such as for industrial environments and outdoor conditions.

One of the first applications for Snowcone is within disaster response, and especially in creating a networked infrastructure when there is a large scale outage of the network infrastructure. Here is a Snowcone:

And a Snowball (with up to 42TB of disk space, an integrated GPU and the opportunity to run several EC2 instances):

And when compared together, we can see the small footprint of the Snowcone:

And a Snowmobile, which can store up to 100 PetaBytes of data storage:

You could just see it in an emergency situation or within a hospital, and providing a backup system to a large scale data infrastructure. It is also tamper-proof and includes a Trusted Platform Module (TPM). This means that the data can be encrypted with the key stored in the TPM, and then using AWS KMS (Key Management Service) for transmitted data back to the Cloud.

Increasingly we will see computing move out of the cloud and back into local devices, and where there is almost zero latency, but where it is all still virtualised. This might include the processing in autonomous vehicles and drones, and where the delays involved in communicating with the cloud could cause issues. Along with this, we may have localised applications that require masses of data to be captured and stored, such as within video broadcasts. This allows for local processing, before the upload into the public cloud. It has also been designed to be robust against harsh environments, such as for industrial environments and outdoor conditions.

One of the first applications for Snowcone is within disaster response, and especially in creating a networked infrastructure when there is a large scale outage of the network infrastructure.

Here is Snowcone:

And here are the Snowmobile and Snowballs in action:

Conclusions

Your company/organisation needs to understand resilience, and data and computational resilience must be part of this plan.