So as expected the event was very much a Google styled event with all of the fixtures and fittings looking like Google had just bought everything that was green, blue, red and yellow from IKEA’s catalogue but it was really well put together and it was really interesting to walk around, there was never a dull moment.
They had candy tubes which dispensed actual jelly beans and this was used to highlight some of Google clouds USPs. They had a full life-size version of chess which was surprisingly popular. Vibrant display stands to draw you in, a pool of ducks obviously, a full arcade complete with retro-styled ssh terminals to allow you to practice your server commands. They also had a complete ‘Dev Zone’ game which was essentially a step-by-step tutorial of some of Google Clouds services. There was also a Lego board with lego artist and a feedback board for questions which every company needs, oh and lastly, a cow, because, why not?
Keynote
So onto the serious stuff! Google opened up the event with their main stage keynote opened by Alan Coad, the managing director of Google Cloud. Alan basically opened up the day and stated the purpose of the event which was basically to bring all the customers of Google Cloud together to share the new innovations that Google has been working on so that we can collectively build better technology which runs in the cloud and move away from stand-alone in-house servers. Which essentially is the essence of cloud technology.
Infrastructure
As the Keynote moved on Alan invited various members of the Google cloud team onto the stage to talk about the new product and capabilities Google was bringing to Cloud. Firstly they mentioned their new N2 and N2D Processors from AMD and Intel which are purpose-built processors designed for Virtual Machines. So unlike the processor in your laptop which is dedicated to your laptop, these processors are designed to be able to run multiple workloads across multiple virtual machines or servers if you prefer. So they are super powerful and these latest versions included multiple performance improvements which will greatly impact the performance of virtual machines in Google’s Cloud.
Next, they introduced, Bring your own IP. So in the past, for any server you spooled up in Google Cloud you would have to assign it a new IP from Google’s inventory, this isn’t great if you’re migrating an existing service and want to keep the same IP. Well, now you can bring your IP address into Google Cloud which is much more seamless.
They also released a handful of other new features such as the acquisition of Cloud Simple, a cloud migration software company. Cloud SQL or SQL Server, New enhancements for SAP infrastructure, support or bare metal Oracle workloads and a Network Intelligence Centre for network troubleshooting and testing in real-time.
Data
Next Google moved the keynote to focus on big data, which is one of the buzzwords of the year. Big data is so important at this point in time and the reason behind it is due to Artificial Intelligence. We are fully aware that Artificial Intelligence is coming and it’s going to have a huge impact on technology and our lives but to make AI work you need data, lots of data. So companies right now are harvesting as much data about anything and everything that they possibly can in order to build AI systems for the future. 5 years ago, companies would often delete or cleanse data and keep only what they needed as to store and query it consumed a lot of resources. Now, however, is the complete opposite as the value of data is so rich and the cost to store and query it in the cloud has reduced significantly that we are now storing huge amounts of data.
Google knows this and so it flexed it’s cloud data warehousing tool, Big Query, by supporting huge data sets of 100’s of petabytes of data and new data engines other than SQL. On their Keynote Google even quoted that one of their customers ran the largest query they have ever seen which resulted in 100 trillion rows of data, to put that into perspective there are 7.8 billion people in the world, there are 1,000 billion in 1 trillion and this query resulted in 100 trillion rows so there are 13 thousand rows of data for every person in that query.
Streaming Data
Google then went on to discuss the processing of live streaming data, so this is data from services such as Netflix and Spotify. This is a type of data collection and storage is particularly tricky but it’s important as 30% of data generated will be real-time by 2025 according to the IDC. For example, this type of data could be tracking something such as the number of times songs have been listened to on Spotify. This is a huge task as the data is consistently changing at the rate of millions of entries per second. Spotify is actually a customer of Google Cloud and this is exactly what they use Google Clouds Data tools for. They use Google Clouds Data Flow tool to track the number of times artists songs have been listened to so that they can pay the artists the correct royalties that they are owed, they also use the exact same system to display a personalised yearly roundup for each Spotify user for the artists and songs that they have listened to the most that year. If you’re a Spotify user, you should have received an email with a link on to your personalised look back already this year.
Lastly one of the biggest announcements for Google’s data tools was ‘Connected Sheets’ this is where you can take data for Big Query, which is a data warehousing tool capable of storing trillions of rows and data and then syncing it to a standard Google Sheet so you can apply filters, graphs, share and analyse the data in a simple and familiar platform. That, to be honest, is quite amazing. Try putting over a million rows into Microsoft Excel and watch it crumble under the strain – that is if your PC will even allow you to that in the first place.
AI
So, moving on, we can’t talk about tech and not mention Artificial Intelligence or AI for short. It’s obvious it’s the future and it’s already becoming established in our everyday lives, the ship has sailed with AI and there’s no looking back. Unsurprisingly, Google is playing a large part in this. They firstly present the Contact Centre AI platform which is now readily available for any company to use and it’s even been integrated with 74% of telephony providers across the world. Contact Centre AI is essentially like having Google’s Assistant working in a call centre, it can independently talk to customers, process and relay information with alarming accuracy and it can also replicate human speech stumblings like when we say ‘erm’ when we’re thinking between sentences and also variance in pitch to sound as if you are just talking to a person.
TPU’s
Okay, so one of the biggest talking points at the whole event was Google showcasing the TPU’s. So, let’s start at the top. A TPU is a Tensor Processing Unit. Tensor is the type of mathematical algorithm required to run Artificial Intelligence-based queries. So a TPU is essentially a computer which has purposefully been built to process tensor algorithms and code and nothing else. Google explained that the reason they have built this is because Moores law has now come to an end. Moores law was a theory which stated that the advancement in processing power doubled every year as we were able to reduce the size of a transistor by half and therefore double the number of transistors on a silicon wafer every 2 years. However, we’ve now reached the physical limit of transistor size and we currently can not make them any smaller. The only way to increase computing power is by multiplying the number of processors in a device or by optimising that device to carry out a very specific task, very well. And that’s exactly what Google has done with their TPU’s. They’ve engineered a processing unit which is perfectly optimised to carry out AI functions and nothing else. So these are not machines that you could run Windows 10 on, for example, they have one job and you pay by the hour to use them for that specific job. Google’s latest TPU board, TPU 3 can process 420 teraflops in a super small footprint.
Google then connects 64 of these TPUs together to create what they call a Pod, which is essentially an incredibly powerful supercomputer than can process AI functions and incredible speeds. These TPU pods are available to use in Google’s cloud platform now and have been for some time, you can rent the usage of a pod in quarter sections, so you could request the processing power of quarter of a pod (or 16 TPU’s) and run you code on that hardware all from a laptop, connected to the internet anywhere in the world. Google claims that these TPU pods can process AI functions faster than any other cloud AI service.
G Suite
Okay so to those that know me they know that I’m a big lover of Google’s G Suite, which is Google’s office tools for business and education. Now, I’m going to make a fairly bold statement here which most people won’t agree with but I’m going to go out on a limb and try to predict the future. So here it is, in 5 years – 8 years time, Microsoft Office won’t be a thing. I truly believe that G Suite is dominating the market and those that haven’t moved over yet are being held back significantly. Microsoft is stuck in the past and can’t bring its users up to date with G Suite. Of course, Google agrees as they stated that independent research claims that companies using G Suite save 21 workdays per employee due to the efficiencies of G Suite and they also have a reduced risk of a data breach by 95%.
Google released a few updates to G Suite which continues to outperform Microsoft’s Office in multiple ways. For example, they have released Smart Compose to Google docs under a Beta. Smart Compose is currently available in Gmail and it’s a helpful feature that auto-suggests the next part of your email based on AI learning of what people tend to write. With this feature now available in Docs, Google will effectively be helping you to write your documents. All powered by Google’s advanced neural networks on their TPUs and delivered to you in a fraction of a second. This is the power of a fully cloud-based office.
Google also released support for Google Assistant with G Suite, so now business users can use Google Assistant capabilities to write emails, book meetings, dial into calls and more, it’s like every employee who uses G Suite gets a free personal assistant with Google.
Support
Lastly, Google finished the talk with a nice hint towards improving their enterprise-level support which is very much needed. Initially, I think Google Cloud went with the stance of here are the tools you need, go and build something and then whatever you build is your responsibility to maintain so we can’t offer support. And to some extent, that’s still true, and the end of the day Google can not become a support team for every application and database built on Google cloud but I think it can do more to help businesses and seems to be making moves in the right direction. General support on server configuration and configuration of the cloud tools will be highly appreciated by system admins and developers across the globe so it’s really refreshing to hear them say this.
Okay so that’s my very top-level highlight summary of my day at Google Next’19, there is so much more I could talk about as I did attend multiple other talks after the keynote which went into more detail about some of the things I have mentioned and more. However, there was so much to take in and so much going on at Next ’19 that it isn’t possible to relay it all back. If you’re interested to learn more about the event and would like to see more detail then I’ll leave a link to the Google Cloud YouTube channel playlist for Next ’19 which has hundreds of videos from the 2 days from talks, to workshops to discussion panels, there really is so much information packed into to 2 short days.