Rohit Jinka

Why IT Employees Don't Want to Return to the Office: 1. 💻 The work gets done remotely. 2. 🚗 No traffic jams. 3. 👨‍👩‍👧‍👦 Better time with family. 4. 🏡 Low cost of living in Tier 2, 3 cities. 5. 💻 The work gets done remotely. 6. 🚫 No need to relocate on job switch. 7. 🕒 Work flexibility enhances productivity. 8. 🌿 Opportunity for healthier living. 9. 😌 Reduced stress and office burnout. 10. 💻 The work gets done remotely. 11. ⚖️ Work-life balance is more achievable. 12. 🚶‍♂️ No more office politics (in person). 13. 🎛️ Control over personal work environment. 14. 👶 Dependent care possible alongside work. 15. 💻 The work gets done remotely. The things that matter to Employers and Businesses are covered in points #5, #10, and #15. Dear Employees, do reshare if these points matter to you too! #RemoteWork #WorkFromHome #ITEmployees #WorkLifeBalance #Productivity

6

Opposition once - Let's ban EVMs Ruling party now - Let's ban Exit polls Opposition once - Ruling party is misleading India Ruling party now - Everyone misleaded us Opposition once - We have to keep our members Ruling party now - Let's add members from somewhere Opposition once - Give us one chance Ruling party now - Let's not take any chance Opposition now - You said 400 is the magic number? Ruling party now - 272 is the magic number and we added some buffer King makers now - We have a BIG WISH list Someone - Please ask media houses who did exit polls to take at least a week break Me - Nobody spoke about stock market #politics #elections #latestnews #trending #newsmedia #india #stockmarket #electionresults #experttalk #justsaying

22

One of the frequently questions asked in the interviews I gave - How to optimize a Spark job? According to me these are the below points which I have answered : 1. Select only the required columns for that specification. Prod table can have 100 to 1000 columns but all the columns are not required. "Select *" is a bad practice. 2. Apply the filter at the beginning while reading the input data. All historical data may not be required each time. Possibly the input table is partitioned and we need only the latest partition to process in the target. 3. Analyse the input data. Maybe the volume of the data distributed unevenly in the files that have been read. Now there are two options either we can 'coalesce' or 'repartition'. I prefer to reparation the data. Coalesce merges the files that are there in the same executor and it can create data skewness. While doing a reparation, if we know the key by which it will be joined in the next step, it is advisable to do reparation with the key as well that the file size comes around 128mb. 4. The target table possibly partitioned. While saving the process data, don't use the saveAsTable API, use insertInto API. It is much faster while writing a large number of data and decreasing the execution time. 5. If a small dataframe is being used multiple times in different parts of the code, cache it. 6. Run some statistics based on the joining key and if the data is skewed, enable AQE. It is one of the powerful features that comes with Spark 3.x and it has many other benefits also. I will write a separate post on it. 7. Try to avoid unnecessary wide transformation. Also avoid cross join. Left join or inner join can be the choice but it depends on the use case. 8. If there is a small table that has been used to join, broadcast it. Keep one executor per node cause this small data will be copied in each executor. Choose graviton EC2 instance for the EMR cluster according to the input data so that the number of executors is equal to the number of nodes. 9. By default spark.sql partitions is 200, but modify the number of it based on the input volume of the data. 10. Keep or select the driver node in such a way that the memory of it is more than the final output. Also, submit production jobs in cluster mode instead of client mode. In cluster mode we don't need to keep track of the driver node. 11. The key by which the join will perform, check if there is NULL for that key. If we don't require NULL key value rows, we can filter out those rows. Else, we can separate it from the join and later do union of that dataframe with the joined dataframe as those NULL keys will not take part in the join. We just need to add columns with NULL values in it before union. 12. Always check the execution plan/DAG. Maybe there are unnecessary repetitive steps that can be optimized. Note- Please add any other optimization techniques that I'm missing and let me know if you have any other thoughts for any of the steps.

144

3 Comments

Reason: why we should do proper error handling in case anything breaks for customers while fetching the data from session/cache etc

17

🚨 Karnataka's Brilliant Plan to Become India's Least Desirable Tech Hub 🚨 In a stroke of absolute genius, Karnataka's government has decided to solve unemployment by making sure no one wants to employ anyone at all. Their groundbreaking strategy? A reservation bill that's about as popular as a mosquito at a nudist colony. The masterminds behind this plan thought, "Hey, you know what our thriving tech industry needs? More red tape!" So they cooked up a bill demanding companies hire Kannadigas like they're collecting Pokemon cards. 50% for management and 70% for the rest - because nothing says "welcome to our global tech hub" quite like "locals only" signs. Unsurprisingly, tech companies reacted as if someone had suggested replacing all computers with abacuses. NASSCOM, ever the party pooper, had the audacity to suggest that companies might actually want to hire based on, oh I don't know, skills? But fear not! The government, in its infinite wisdom, has now put this brilliant plan on hold. Apparently, they've realized that turning Bangalore into a ghost town might not be great for the economy. Who knew? Meanwhile, real estate agents are salivating at the thought of all those empty apartments they could soon be not renting out. After all, nothing boosts a rental market quite like an exodus of skilled workers! In the end, Karnataka faces the thrilling challenge of satisfying local job demands without completely torpedoing its economy. It's like trying to juggle flaming torches while riding a unicycle - what could possibly go wrong? Stay tuned for the next episode of "How to Sabotage Your State's Economy in One Easy Step!" Brought to you by the Karnataka government - where common sense is optional, but reservations are mandatory! #namma #Bengaluru #Reservation #economy #Karnataka #jobs #humour

2

🔹Should You 🫵Buy Apple Products in India? 🤔 It depends on the math. 📊 Someone shared this interesting perspective with me online, and it’s worth considering. 🌐 For example, it takes just 🔹#21days🔹 of savings for a USA citizen to buy an iPhone 14 Pro Max, while it takes 🔹#218days🔹 for an Indian citizen to save up for the same phone. 😲💸 Before starting another EMI, think about this carefully. 💭💡 #Apple #iPhone16 #FinancialPlanning #ThinkBeforeYouBuy #EMI #Savings #TechTalk #IndiaVsUSA #SmartShopping

3

2 Comments

🔔 Karnataka’s Political Landscape Unveiled 🔔 🔻 Agriculture Budget Reduced by 38%: Farmers suffer as support dwindles. 🔻 Irrigation Budget Slashed by 17%: Fields thirst, crops hang in the balance. ⛔️ PM KISAN Samman Scheme Withdrawn: Monthly sustenance halted. 🚫 Jeevan Jyothi Insurance Scheme Abandoned: Safety net torn. 🚫 Raita Vidya Nidhi Scheme Neglected: Scholarships deferred. 🚫 Bhoo Siri Scheme Discontinued: Transparency relegated. 🚫 Raita Shakti Scheme Left in Limbo: Backbone weakened. ❌ Kaveri Water Released to Tamil Nadu: Political chess, adverse orders. Congress, your audacity astounds. DK Shivakumar, your vacuous arguments crumble. Explain these decisions. Let truth emerge. 🌟 Insight: Leadership thrives on action, not slogans. Karnataka awaits true stewards of progress. 🌱🇮🇳

1

#JioCare Beware of buying any products of Jio. they have so many hidden clauses in refund. I had bought a JioAirfiber and paid the whole amount. 1. The technical team did not turn up the day when it was supposed to be installed. 2. Manager Arvind asked for full payment of the Jio Fiber and promised that it will be installed the very same day 3. I called up Customer care and was given standard answer that technical team will come for installation. 4. Technical team came after 3 days and informed that they cannot install due to technical issue. 5. They immediately left and I was told to contact customer support for refund. 6. On contacting them one team said you will receive in 7 days. every time I called I was given different answers. 7. i purchased the Jio service on 30 March and have reported to Jio Appellate body but nothing is working. Always buy from local service provider, they are best. Boycott Jio. reference: Service Request number SR00001WOQ3K

1

1 Comment