Igeometry Podcast

Backend database applications relay on good storage systems for performance, durability and low latency. SSDs have been a great invention that changed the storage game compared to mechanical drives. However, SSDs came with their own sets of problems which can cause database reads and writes to plummet after a while. This episode discusses the current limitations of SSDs include garbage collection, wear leveling, over provisioning and how zoned name spaces “may” fix this. 0:00 Intro 1:40 Block 4:00 Logical Block Addressing 6:00 Flash Translation Layer 6:50 DRAM 8:50 Erase unit 11:11 Namespace 12:00 Summary 13:40 Garbage collection 16:49 Over provisioning 18:00 Write amplification 21:28 Wear leveling Resources https://www.guru3d.com/news-story/new-malware-bypasses-security-measures-by-using-ssd-over-provisioning.html https://www.snia.org/educational-library/zoned-namespaces-zns-ssds-disrupting-storage-industry-2020 https://www.youtube.com/watch?v=cbX3P56Jp0o&feature=emb_title --- Support this podcast: ht

Google linux boxes have over 16 NVMe SSD PCIe Express drives.When a shutdown signal is sent to linux, the OS iterate through each NVMe and send synchronous request to shutdown that takes 4.5 seconds. This adds up to over a minute to achieve a reboot. Google patches linux with asynchronous Shutdown APIs    Resources https://www.phoronix.com/forums/forum/phoronix/latest-phoronix-articles/1316262-google-has-a-problem-with-linux-server-reboots-too-slow-due-to-too-many-nvme-drives  patch https://lore.kernel.org/lkml/20220328230008.3587975-1-tansuresh@google.com/  Sync vs async  https://www.youtube.com/results?search_query=asynchronous+vs+synchronous+hussein --- Support this podcast: https://anchor.fm/hnasr/support

Hash tables are effective for caching, database joins, sets to check if something is in a list and even load balancing, partitioning and sharding and many other applications. Most programming languages support hash tables. However they don’t come without their cost and limitations lets discuss this. 0:00 Intro 1:50 Arrays 3:50 CPU Cost (NUMA/M1 Ultra) 6:50 Hash Tables 10:00 Hash Join 16:00 Cost of Hash Tables 20:00 Remapping Cost Hash Tables 22:30 Consistent hashing --- Support this podcast: https://anchor.fm/hnasr/support

Aggregate functions like Count, max, min, avg performance really depends on how did you tune your database for that kind of workload. Let us discuss this. 0:00 Intro 1:22 SELECT COUNT(*) 4:30 SELECT AVG(A) 5:15 SELECT MAX(A) 8:00 Best case scenario 11:30 Clustering 14:00 Clustering Sequential Writes 17:19 Clustering Random Writes 20:30 Summary --- Support this podcast: https://anchor.fm/hnasr/support

99% of phishing attacks can be avoided by looking at the URL. However this secuirty researcher proves that not enough.   https://mrd0x.com/browser-in-the-browser-phishing-attack/ --- Support this podcast: https://anchor.fm/hnasr/support

On March 8 2022 Spotify and Discord experienced an outage latest 2-3 hours. The reason was a configuration on the xDS formats on Google Traffic Director. Let us discuss how this change caused the outage and what Spotify did to mitigate that outage without relying on Google restoring the service back up. Resources Spotify outage https://engineering.atspotify.com/2022/03/incident-report-spotify-outage-on-march-8/ Google Cloud outage https://status.cloud.google.com/incidents/LuGcJVjNTeC5Sb9pSJ9o Envoy xDS https://blog.envoyproxy.io/the-universal-data-plane-api-d15cec7a Microservices scaling with common sense https://www.youtube.com/watch?v=NsIeAV5aFLE CARDS 4:36 Miicros https://www.youtube.com/watch?v=NsIeAV5aFLE 9:30 Spotify Hermes https://www.youtube.com/watch?v=fMq3IpPE3TU&t=21s 0:00 Intro 2:00 Spotify Outage 3:30 Microservices 6:10 Service Discovery 10:00 Spotify Quick Workaround 12:15 Google Traffic Director Outage --- Support this podcast: https://anchor.fm/hnasr/support

The Russia-Ukraine Cyberwar has reached a peak this past week. In this video I explain the multiple ways a Denial of Service attacks can happen and how to protect against them Intro 0:00 What is DOS? 1:00 Long running requests 1:42 Crash backend processes 3:30 Exhaust Max connections 4:25 Large response (web traffic) 5:42 Lots of requests 8:27 Complex request (Regex bug) 9:57 Prevention 12:27 --- Support this podcast: https://anchor.fm/hnasr/support

 Fundamentals of Database Engineering udemy course (link redirects to udemy with coupon) https://database.husseinnasser.com Chrome, Edge and Firefox are approaching version 100. that version is 3 digit version where most websites only checked two digits. You can imagine what can go wrong. Lets discuss --- Support this podcast: https://anchor.fm/hnasr/support

On 2/22/2022 Slack experienced a 3 hours outage from 6AM to 9AM PST because of a configuration change. That change lead to increase load to their database infrastructure which causes queries to starve. Lets discuss https://status.slack.com/2022-02-22 --- Support this podcast: https://anchor.fm/hnasr/support

It looks like there is demand for bringing QUIC and HTTP/3 to NodeJS, someone donated $1000 dollar to make it happen. However, what are the use cases for supporting HTTP/3 in Nodejs? I talk about whether HTTP/3 should come to NodeJS in this video. https://rysolv.com/issues/detail/863986ce-c8d5-466d-abf4-476d3177452d --- Support this podcast: https://anchor.fm/hnasr/support

Five major Canadian banks went offline for hours blocking access to online and mobile banking as well as e-transfers for customers. There are no reports of what might have caused the outage so In this video we explain what might have caused it.   https://www.bleepingcomputer.com/news/security/canadas-major-banks-go-offline-in-mysterious-hours-long-outage/ --- Support this podcast: https://anchor.fm/hnasr/support

NVMe changed the game of fast SSD storage. The spec is being extended to support networked storage, I have some doubts that would like to discuss in this video.  Fundamentals of Database Engineering udemy course (link redirects to udemy with coupon) https://database.husseinnasser.com Introduction to NGINX  (link redirects to udemy with coupon) https://nginx.husseinnasser.com Python on the Backend (link redirects to udemy with coupon) https://nginx.husseinnasser.com --- Support this podcast: https://anchor.fm/hnasr/support

I stumbled upon a research calculating how big of quantum computer required to break bitcoin public key encryption. It will take them 300 million quantum bits to find the corresponding private key given a public key. That wasn’t really the most instructive thing really from that research. The interesting thing is how bitcoin design carefully hides information about the original owner which makes breaking it really challenging. In this episode I discuss what does it take to break bitcoin encryption. 0:00 Intro 3:00 Public Key Cryptography 6:30 Bitcoin Keys ( Private key/public key and bitcoin address) 8:18 Why do we need a Bitcoin address 11:13 How minors verify transactions 17:30 When is the bitcoin public key visible? 19:12 The Wallet 20:47 What does it take to break bitcoin? -find wallet.dat file which contain all the keys, some upload that online -obtain public key from digital sign during a transaction posting. If someone used that public key , obtained the private key, then used it to generate a new rans

NodeJS introduces importing network based modules in their latest release. Importing network based modules using `https:` and `http:` is supported under the `--experimental-network-imports` flag. This allows web browser-like imports to work in Node.js with a few differences due to application stability and security concerns that are different when running in a privileged environment instead of a browser sandbox.  Import HTTPS https://github.com/nodejs/node/pull/36328/files --- Support this podcast: https://anchor.fm/hnasr/support

I asked a question on my community post   What port does your computer connect to when you visit google.com on your browser for the very first time?  The answer really depends on the network configuration and that's what I discuss in this video. The answer could be HTTP port 80, HTTPS port 443, DNS port 53 or other answers that aren't even listed here including DNS over HTTPS and DNS over TLS. Lets discuss --- Support this podcast: https://anchor.fm/hnasr/support

In this episode of the backend engineering show I discuss the ramification of index page splits which results in fragmented index yielding slow query performance when using indexes. I go through what a page is, how a page is read and written in the database and finally discuss the current solutions to address this problem such as reindexing, fill factor and choosing a good index key type. Get my Fundamentals of Database Engineering Udemy course https://database.husseinnasser.com 0:00 Intro 1:50 What is a Page? 5:00 How to Read and Write a page 8:45 Batching Writes and WAL 11:45 The Order of Content in the Page 16:00 Page Splits 21:30 Solving Index Fragmentation 28:30 Summary --- Support this podcast: https://anchor.fm/hnasr/support

Multitenancy refers to placing two or more clients, customers or tenants so they share a single resource. The idea behind Multitenancy is to save on resources instead of giving each client its own resource (storage, app, memory, network etc..). You slice each part and give it to dedicated client. The number one rule is those tenants shouldn’t be aware of the existence of other tenants, they need to be completed isolated, no one tenant can read or influence the other’s tenant share. If this is confusing don’t worry I’ll go through examples in this episode of the backend engineering show. 0:00 Intro 2:50 Ssd Multitenancy 6:45 Database Multitenancy 10:14 Networking Multitenancy (SDN) 16:55 VMs and Containers Multitenancy 19:00 Application Multitenancy --- Support this podcast: https://anchor.fm/hnasr/support

Scientists at the Physics and Engineering Department of the UK’s Lancaster University published a new paper describing ULTRARAM. A Low-Energy, High-Endurance, Compound-Semiconductor Memory on Silicon. I think this tech is a game changer but I still have my doubts. Let us discuss 0:00 Intro 0:30 RAM 4:00 SSD 11:00 UltraRAM 17:30 Doubts * RAM (Dynamic RAM) * Fast access with capacitors * random access * requires power to be refreshed * Flash NAND (SSD) * Pros fast random access (FTL) * Block storage * Requires high voltage (20V to erase/ 7.5 to program) * Low durability endurance which then (write amplification + garbage collection ) * UltraRAM uses a new novel approach to the memory cell design that only need 2.5 V to program/erase, low power which saves the memory cell lifetime https://www.sciencedirect.com/topics/engineering/fowler-nordheim-tunnelling * High endurance program/erase last longer * High retention * GC/WA not required * Limitations: still we know very little h

enom my DNS registrar had an outage which took down my site husseinnasser.com lets discuss   0:00 Intro  5:00 How DNS Works 12:00 Outage report 16:00 Will I migrate my DNS? 21:40 Does decentralize mean anything anymore? --- Support this podcast: https://anchor.fm/hnasr/support

In this episode of the backend engineering show, I’ll discuss the low code movements and their pros and cons. 0:00 Intro 2:00 Evolution of Languages 8:40 Low Code Movement 10:00 Side Effect of Low Code 13:00 Leaky Abstractions Fundamentals of Database Engineering udemy course (link redirects to udemy with coupon) https://database.husseinnasser.com Introduction to NGINX (link redirects to udemy with coupon) https://nginx.husseinnasser.com Python on the Backend (link redirects to udemy with coupon) https://nginx.husseinnasser.com --- Support this podcast: https://anchor.fm/hnasr/support