Thread by @TelcoDR

Public Cloud is coming to all parts of #telco.

This is the @totogi charging engine chapter.

Simply put: the key to Totogi's amazing performance is the public cloud.

@totogi is the first charging engine to be designed and engineered from the ground up assuming public cloud. (In this case, @awscloud is the public cloud)

Most charging engines that are being ported to the public cloud, which claim to be "cloud-native," are based on a tech stack of Kubernetes and a relational db. This is a HUGE liability compared to what @totogi designed.

By writing to the core of AWS services, @totogi can be 10x more scalable, resilient, AND 80% cheaper.

Many will argue over the words cloud vs. cloud-native.

Here is a hint: Look at pricing to find out who truly designed for the cloud. 🧐

Let's take 3 parts of the @awscloud stack as an example:

- Ubiquitous data centers connected with high-speed networks
- Graviton (ARM) processors
- DynamoDB

Ubiquitous data centers are one of the most notable features of hyper-scale, cloud vendors. They are everywhere. They enable the great uptime and disaster recovery needed for telco.

They are one of the strongest counterpoints to having your own on-premise charging engine. One built native for the cloud is more secure, more resilient, has more uptime, and is more scalable because of the data center footprint of the hyperscalers.

Charging engines are processing monsters. They are chewing through CPU to process all those transactions. 👹

Instead of using Intel/AMD x86 processors, @totogi built its system on Graviton2/ARM processors because they are the fastest processors available and also provide 40% better price/performance.

Graviton2/ARM is the fastest and cheapest - just like Totogi.

You can't get these chips for your on-premise data center and if you're just porting your on-premise app to Intel/x86 in the cloud, you aren't taking advantage of it either.

(Oh yeah - Graviton is just getting started. Look at what ARM has announced that will be showing up soon)

I can't wait until these new instances are available and I just have to click a button to get HUGE performance increases in the @totogi engine.

Have fun with your 5-year depreciation cycle on those Intel boxes! 👋🏼 👋🏼

And finally: @dynamodb.

DynamoDB may be the biggest, most controversial of the technical decisions for @totogi.

All charging engines are run on relational DBs (think Oracle) or niche in-memory DBs (e.g. Oracle's latest telco charging benchmark uses one, or Matrixx has hyped theirs for years).

These DBs are HUGE liabilities which is why legacy chargers have limitations around performance.

Everything around a legacy charging engine has to be super tuned and optimized. They always require more hardware and upgrades. You can't have that rate plan because the legacy charging engine can't handle the performance needed.

By using @dynamodb, we throw all those limitations out. DynamoDB can scale to enormous numbers. Most charging vendors talk about the largest CSPs that they can support (e.g we can handle the 20m subs of a tier 2)..

Totogi, because of @dynamodb, talks in terms of how many COUNTRIES it can support (e.g. we can handle all 1.1b subs in India, as well as the 750m of the EU at the same time)

For a few years now, AWS has been touting @dynamodb and single-table design as a better cloud solution than a relational db. They rewrote all their internal systems to get off Oracle. They pushed the controversial concept of single-table design.

As startups should, @totogi saw the opportunity to build a system that no one else could match. One that could upend the big charging players, like Amdocs and Ericsson.

@totogi made a core technical decision that was so valuable, it would FORCE the incumbents to REWRITE their systems from the ground up.

The only problem? No one believed you could build a single-table designed, DynamoDB based charging engine.

(Just ask all the incumbents. They didn't make it work and they will be happy to retroactively justify to you why it still doesn't work - our demo at MWC notwithstanding.)

Obviously, they haven't seen our #MWC21 demonstration yet.

So, the @totogi engineers got out the coffee, put on the headphones, and went to non-stop work. ☕️🎧👩‍💻

And they delivered! Get excited to see the results at #MWC21. 🤩

In a world where a "big time" dinosaur like Oracle releases a whitepaper benchmark that is 130,000 tps, and a "no-name" player like Matrixx releases a benchmark that is 200k tps, @totogi is demonstrating 1 MILLION tps.

@totogi is committed to 2x tps every 6 months until it can concurrently handle every subscriber on planet Earth. (We'll be talking to @SpaceX about adding Mars soon...)

With all 3 cloud technologies together - you will see @totogi running across 6 data centers on 2 continents.

And here's the kicker:

@totogi has the ability to lose an ENTIRE datacenter without dropping a single transaction and the loss of only 1.3 seconds of missed charges if an entire continent goes away - all while running 8x more throughput than Oracle's best benchmark.

While the @totogi engineering team will be taking a bow, they also acknowledge the shoulders of the giant (@awscloud) they built on.

Thanks to @jeffbarr @houlihan_rick @alexbdebrie @QuinnyPig @forrestbrazeal @mcbride_aws

Public cloud is coming to all parts of telco - this is the charging chapter. Want to see it in action? Visit #CloudCity at #MWC21!