About this episode
Welcome to episode 247 of the CloudPod Podcast – where the forecast is always cloudy! Pepperidge Farm remembers – and now so does ChatGPT! Today on the pod we’re talking about the new “memory” function in ChatGPT, secrets over at OCI, and Firehose dropping Kinesis like its HOT. Plus plenty of other Cloud and AI news to get you through the week. Let’s get started! Titles we almost went with this week: I Don’t Think Anyone Wants to be “Good Enough” in AI ㊙️Oracle Can Rotate All My Secrets Amazon Data Firehose – Not Without Kinesis A big thanks to this week’s sponsor: We’re sponsorless this week! Interested in sponsoring us and having access to a very specialized and targeted market? We’d love to talk to you. Send us an email or hit us up on our Slack Channel. Follow Up 00:57 C2C Event Recently Justin was down at a 2gather event Google’s Cloud headquarters near Moffett Field in Sunnyvale. So to those new listeners who heard Justin there and just couldn’t get enough, welcome! We’re happy to have you. Want to see what events are coming up, and hopefully near you? Check out the lineup here . General News 08:25 Why companies are leaving the cloud A recent study by Citrix , is saying that 25% of organizations in the UK have already moved half or more of their cloud-based workloads back to on-premises infrastructures. The survey questioned 350 IT leaders on their current approaches to cloud computing. 93% of them had been involved in a cloud repatriation project in the last three years. Surveyed said their reasons for moving from the Security Issues, High Project Expectations and unmet expectations, with most saying the cost was the biggest motivator, which definitely makes sense to us. In general this isn’t my experience when talking to listeners, or folks at the recent C2C event; there’s always a few companies that probably shouldn’t have moved to the cloud in the first place, but those numbers don’t pan out to us in who we’re talking to. We’re interested in listener feedback here – have any of you been involved in a repatriation project? 09:55 Ryan – “ I think it’s kind of the same thing that happened in reverse a few years ago, where it’s like all the companies are moving to the cloud. The same reports were, you know, 50 % of companies are moving other entire workloads into the cloud. And now it’s sort of the pendulum swinging the other way.” AI is Going Great (or how ML Makes all Its Money) – ChatGPT gets Reveries 12:37 Memory and new controls for ChatGPT ChatGPT is adding a new “memory” feature; “remembering” allows you to ask the bot to remember things you have chatted about with ChatGPT in the past. So things like you love to travel, you have a daughter, etc. It’s as simple as asking ChatGpt to remember something while you’re chatting with it. If you don’t want memory you can enable “temporary” chat. If you are creating GPTs’s, you can also setup memories for those as well. 15:04 Ryan – “ I think a lot of people initially trained a lot of models to get that level of customization, right? So they are building their own models based on that, which is super expensive. And so now this is sort of an option to get, you know, this is sort of in the middle, right? It’s where you want, you want some of these things to be sort of general biased things that you’re setting, but then you can use just the model as is after that, which is great. 15:31 Cohere For AI Launches Aya, an LLM Covering More Than 100 Languages Cohere for AI, Cohere’s non-profit research labs, are excited to announce a new state-of-the-art, open-source, massively multilingual, generative LLM covering 101 different languages Big improvements include the large language support and cultural relevance, which is very important to using multi-lingual models 16:07 Justin- “ A lot of people are looking at LLM for things like localization support and translating tweets and different things, different languages so people have that access and you need to make sure it’s being culturally relevant. Or else you’re going to end up in a good PR nightmare, which is not great.” AWS 18:07 Knowledge Bases for Amazon Bedrock now supports Amazon Aurora PostgreSQL and Cohere embedding models Knowledge Bases for Bedrock allows you to securely connect foundational models in Bedrock to your company data for Retrieval Augmented Generation (RAG). They have expanded that to support Amazon Aurora PostgreSQL , the vector engine for OpenSearch Serverless , Pinecone , and Redis Enterprise Cloud . 19:50 Justin – “S o basically there’s a picture of the flow of this in the document and basically there’s a user query. It goes to basically the embedded retrieval augmentation and gives you an embedding model that then generates the embeddings that make sense. And then basically it takes that, applies it to the vector store to retrieve similar documents, then uses those similar documents to search the foundational model to basically augment the two together. And then that’s what responds back to you as the user.” 21:46 AWS positioned highest in execution in the 2023 Gartner Magic Quadrant for Cloud Database Management Systems AWS was named the leader in Magic Quadrant for Cloud Database Systems . It is great if you care about the number of options, including Elasticache and RDS for Db2. Also a factor was the ability to do vector searches in the DB of your choice, Zero-ETL integrations and Generative Power AI powering our data services. 23:53 Ryan – “I’m surprised Amazon’s not closing on Google in terms of being more visionary; with a lot of the enhancements that they’ve put out in the last few years; I do get Google leading in the visionary space and Amazon leading in the ability to execute.” 26:49 Introducing Amazon Data Firehose, formerly known as Amazon Kinesis Data Firehose AWS is renaming Amazon Kinesis Data Firehose to Amazon Data Firehose . Amazon Data Firehose is the easiest way to capture, transform, and deliver data streams into Amazon S3, Amazon redshift, Amazon Opensearch Service, Splunk, snowflake, and other 3rd party analytics services. Let’s be real; the Kinesis name was definitely a mistake as it can support more than just that. 30:50 Justin – “ I definitely think Kinesis could have some legs with it. If it could get some of the things that Kafka has for on -prem and like if they’ve truly supported hybrid properly, I think Kinesis could have a lot more traction, but I think they’ve so limited what they’ve done for supporting Kinesis on -premise to really just, you know, their, their big iron and primus appliances. I think that’s where it limits their ability to really compete with Kafka.” GCP 31:59 For Google, ‘Good Enough’ Gemini AI Could Be Good Enough to Win Early reviews for Googles Gemini Ultra model are pretty good and make it appear that it will be good enough to win Ultra is much faster, less wordy and less bland in answers than the paid version of Chat GPT. It has even done a good job with creative storytelling However, it falls short in a number of areas, most noticeably coding and reasoning problems. 36:08 Justin – “ There’s some cool stuff coming; it’s showing up everywhere. Again, while I think it’s changing the world in many ways, and I think it’s fundamentally changing some jobs (like copywriters ) I think it’s still a bit overhyped. 37:42 Announcing the general availability of Network Function Optimizer for GKE Enterprise Google is announcing the new Network Function Optimizer is generally available for GKE Enterprise , the premium edition of Google K8 engine. As part of the GKE Enterprise, network function optimizer delivers the enterprise scale nad high data plane performance for containerized applications that Google’s customers have been looking for, including the functionality that have been developed as part of the multi-network K8 enhancement proposal and our Multi-Network , new level of multi-tenancy presentation into the K8 community. Some key benefits: Extending multi-network capabilities to PODs that run on the nodes. With multi-network support for Pods, you can enable multiple interfaces on nodes and pods in the GKE cluster, allowing for data-plane and control-plane separation. Delivering a high-performance data plane natively in software that is comparable to those assisted by hardware, simplifying workload scheduling on the pod and removing underlying hardware/Nic dependency. 38:30 Justin- “ This is the first time I’ve seen them talk about GKE Enterprise in a hot minute. Wasn’t that one of the things we talked about with them at Next is that they were gonna start either deprecating this name and moving just to Anthos or Anthos was gonna deprecate into GKE Enterprise. Then nothing’s really happened in that space…I was really looking back to the article, bringing together the best of GKE and Anthos into an integrated intuitive container platform with a unified console experience. It’s like, to me, it sounds like they’re combining, but yeah, so it seems like they had a vision and then they sort of forgot about it. Or maybe, maybe they just haven’t finished development and then we just haven’t heard much because November wasn’t that far ago.” Azure 45:01 New data and AI solutions in Microsoft Cloud for Sustainability help move organizations from pledges to progress Today, we’re introducing new data and AI solutions for Microsoft Cloud for Sustainability that provide capabilities for organizations that need to progress in their sustainability journey. These include faster environmental, social and governance (ESG) data analytics and insights, AI insights, an AI assistant to help accelerate impactful decision-making and reporting, and other advanced capabilities Now in preview, sustainability data solutions in Microsoft Fabric allow organizations to accelerate their time to insights and sustainability progress by providing out-of-the-box ESG data models, connectors and reporting. By connecting your ESG data to the fabric, you can turn volumes of sustainability data into meaningful insights and progress. 46:37 Ryan – “ The irony in all this for me is that out of all the clouds, the one to get your sustainability data out of that’s the hardest is Azure. Everywhere else has this managed thing; I can go directly to a dashboard and I get that number and I can export a report. With Azure, I got to set up this Power Blink to this app thing that links to a template in a database, which I then have to authorize at the main tenant level of my Microsoft…and I’m laughing at this; make me jump through 12 hoops to get it.” 49:48 Azure Elastic SAN is now generally available GA of Azure Elastic San , the industry’s first fully managed and cloud-native storage area network (SAN) offering that simplifies deploying, scaling, managing and configuring a SAN in the cloud. Azure Elastic SAN responds to the vital need for seamless migration of extensive SAN environments to the cloud, bringing a new level of efficiency and ease. This enterprise-class offering stands out by adopting a SAN-like resource hierarchy, provisioning resources at the appliance level, and dynamically distributing these resources to meet the demands of diverse workloads across databases, virtual desktop infrastructure, and business applications. Investigate performance and capacity metrics with Azure Monitor Metrics Prevent incidents due to misconfiguration with the help of Azure Policy 50:30 Justin- “Can you think of anything less cloudy than a SAN?” Oracle 52:55 Automate secret generation and rotation with OCI Secret Management OCI Secret management offers a robust and secure solution for storing, managing and accessing these secrets. It provides centralized storage protected by HSM and granular access control to help ensure the security and integrity of your secret, and now you get automatic Secret Generation and rotation capabilities. Automatic secret generation can create passwords, SSH keys, random bytes, and templatization capabilities that allow you to do things like store JSON blobs with placeholders for secrets that are automatically generated for you. With automatic rotation, you can set intervals from 1-12 months. This feature integrates with the autonomous database and function services, allowing seamless rotation of secrets in the Autonomous database or function code. 53:33 Ryan – “I think the biggest secret they’re keeping is who are the Oracle Cloud customers?” Continuing our Cloud Journey Series Talks After Show 54:50 How do subsea cables work? So we recently talked about some new subsea cables, and the emerging tech brew blog wrote a story briefly describing a few things about subsea cables work, so we figured we’d share! Most cables have 16 slim fiber-optic strands that transmit the data, surrounded by a layer of copper armoring to protect and stabilize the strands. This is then encased in a polyethylene jacket. The lightweight cable, typically no bigger than the circumference of your thumb, runs through the deepest parts of the ocean. As the cable reaches shallow waters, it is further armored with additional casings up to 2 inches in diameter. The cables are laid by ship with 1000’s of miles of cable coiled in the bowels being fed off the back of the ship much like unfurling rope. The ship has a plow that creates a trough in the bed of the ocean for the cable, and the underwater currents eventually bury it in the sand. Periodically the cable will run through a housing that is designed to last 25 years to amplify the data on its journey. This process hasn’t changed since the telegraph – only the amount of data has increased. There are roughly 500 subsea cables that traverse the globe. Closing And that is the week in the cloud! Just a reminder – if you’re interested in joining us as a sponsor, let us know ! Check out our website, the home of the Cloud Pod where you can join our newsletter, slack team, send feedback or ask questions at theCloud Pod.net or tweet at us with hashtag #theCloud Pod