AWS This Week: New Amazon EC2 P4 instance, Redshift supports Lambda UDFs and tokenization, and more!

Okay, hello Cloud Gurus, how y'all doing? I'm Scott Pletcher and
this week on AWS This Week, Redshift gets a nifty new
feature, application load
balancers get GRPC supports, and we get our hands on a
massive new EC2 instance. Plus, we'll announce the coveted guru
of the week award. Stick around. [Inaudible]. Good news for all you web
application developers out there. Application load balancers are now
capable of supporting end to end HTTP/2 and gRPC. HTTP/2 has been around for a while
and it's an improvement over HTTP/1.1 by changing the way the
data is transported between
the client and the server, resulting usually in better performance. Better performance means
faster web applications. gRPC is an open-source for remote
procedure call framework that uses HTTP/2 and has become quite popular
among application developers. You can also use ALB's content
routing features to inspect the gRPC calls and route them to specific
target groups based on the service and method request. Recently, AWS announced that you can now use Lambda
functions as user defined functions, right in Redshift.

User
defined functions, or UDS, are little chunks of custom code that
you can embed directly in your Redshift SQL statements. Now, because
these are Lambda functions, you can use any language
that Lambda supports, even bring your own custom runtime. One such use case for this
functionality is tokenization. Tokenization is a strategy by which you
can obscure and dynamically replaced sensitive data with some other
nonsensitive value as a sort of stand in. You could, for example, use a Lambda UDL and your insert statement
that dynamically calls out to some other system and exchanges a credit
card number for a harmless token, thus keeping payment card numbers
out of your Redshift database and therefore outside the scope of PCI audits. This week, AWS announced a new
EC2 instance of massive scale. The P4d instance is designed for
heavy duty machine learning and high-performance computing
applications. Now, the new instance only comes in one
size right now, sporting 96 VCPUs, 1.1 terabytes of system memory,
8 terabytes of SSD storage, and is capable of 400 gigabit
per second network throughput. The instance also contains
8 NVIDIA A100 tensor core GPU's with 320 gigabytes of GPU memory.

Now these monster beasts are
only available in US-East-1
and US-West-2 right now and will set you back just
under $33 per hour on demand, but they are also available
as spot instances, reserved instances and part of
savings plans. According to AWS, the P4d instance delivers up to
2.5 times better performance at up to 60% lower cost than the
equivalent P3 resources on machine learning training tasks. And no, the math still doesn't work out
for profitable Bitcoin mining.

Yes, the time is upon us to reveal this
week's guru of the week. Congratulations, go to Ian Evans, an enterprise
architect from San Jose, California. A lovely swag pack is on
its way to you as I speak. If you'd like to have a go at this
week's guru of the week challenge, please check out our forum
link below. That, my friends, is all the AWS news that's
fit to print this week. For more AWS and A Cloud Guru goodness, be sure to click those
subscribe and notify buttons, or just give us a thumbs up and drop us
a line down in the comments.

Stay safe, take care of one another, and
keep being awesome cloud gurus..

You May Also Like