Google Cloud Product Fundamentals & Managing Change when Moving to Google Cloud

M1:
Outsource infrastructure because management and maintenance is a cost. Enabling technology is the VM. Shift cost from capital expenditure heavy to operational heavy. Google has 40% of internet traffic passing through it. On-prem private clouds can work with cloud, like hybrid or multi cloud, enabling technology is Anthos. Google handles lower layers of security, hardware, software, storage, identity, network, and operations. Depending on your managed to self serve infrastructural choices, you will have the ability to pick products that do what Google offers in managed products. GCP resource hierarchy, project resources, project, folders, organization. Permissions are transitive, the more generous policy take effect. Who can do what on which resource. Primitive roles: owner, editor, viewer, (and billing). Pre-defined roles, custom roles. Orgs use least privilege roles. You can use groups to structure permissions. Google offers Compute, Storage, Big Data, Machine Learning. Compute: infrastructure, platform, serverless functions. Storage: based on type, structured and unstructured. Big Data: processing tools and orchestrating tools. ML: pre-trained models and ability to train your own.

M2:
Need to modernize your applications to take advantage of cloud. Think about storage in conjunction with development. Lift and shift, change apps before move, change apps after move, invent greenfield, invent brownfield (replace existing cloud app), move applications without any changes. Focus on building, not infrastructure and maintenance, else people leave. Think micro-service over monolith, CI/CD for faster delivery, and rollback. App Engine manages infra for you, can run multiple versions of the app. Data is structured (string, number, ref) and unstructured (audio, word files, blobs). Database to hold data and answer questions via queries. Cloud SQL a RDBMS for website serving, operational apps, and BI. Cloud Spanner for data that is copied across regions, great for availability, queries are consistent. BigQuery as storage service and fast analysis service. Multi-regional storage is great for disaster planning, regional is local use and is great for analytics and performance/throughput because of compute co-locality. Cloud Storage Nearline for data accessed once per month like file storage and backups, and Coldline for data accessed once per year like archives or disaster recovery backups.

M3:
Most data is backwards looking, you need to make predictions for forward looking trends like predicting seasonal and other operation specific data points. ML is standard algos to get predictive insights from data to make repeated decisions. ML is branch of AI. Once ML is trained with input and correct output (label), it is a model. Model is only good as the data is, focus on coverage, cleanliness, and completeness. Predictive analytics is future, great for repeated decisions. Growth of ML models used in products is exponential. Google has a wide range of AI products for exact needs, AI hub, prebuilt models, tensor hardware for accelerated training.

M4:
G Suite is the collaboratively working tool. Enable move with speed (agile, realtime), informed decisions, collaborative culture. Use the explore tool to get insights via AI. G Suite allows for cross org sharing and messaging.

M5:
Managing cloud costs like planning and budgeting, monitoring and controlling, accurately forecasting. Avoid spiralling costs, underestimating cloud spend, as team scales you need more granular and aggregate views. Total cost of ownership is the assessment of all layers within infrastructure and other associated costs across the business over time like hardware, software, management, support, communications, user expenses, service downtime, training, and other productivity losses. Think about value gained over time, and indirect value. Visibility to who, finance, IT? Develop accountability with clear ownership, and controls. Create budgets and spend notifications.

Managing Change when Moving to Google Cloud

70% of large and complex programs fail to achieve goals. First focus on leadership and how they can support the transition, then focus on collaboration to break down business and technology silos. Then focus on capabilities and skills. Most people resist change, there is a spectrum of risk takers.

Framework for adopting cloud (Venn diagram): Tech (networking, resource manegement) (external experience, up skilling) people (people ops, behaviours, communication) (sponsorship, teamwork) process (cost control, incident mgmt, instrumentation) (architecture, infra as code, CI/CD) ((identity & access, data mgmt)). We need to continuously learn, lead effectively, scale, and secure. 90% of our behaviour is influenced by our values, beliefs, and habits.

Break down the maturity journey into tactical, strategic, transformational. Tactical's goal is to reduce cost of discrete systems, quick wins, not scalable, short-term. Strategic's goal is to be broad in vision governing individual workloads based on future needs and scale, embracing change, mid-term goal. Transformational has cloud operations functioning smoothly and are integrating data and insights gathered from working in the cloud, long term, use ML for prescriptive analytics and IT is not a cost center but a part of business.

Organizational change is not linear: move first then change for slow adoption, change before move for aggressive adoption, just move is popular for infrastructure layer changes and not necessarily for processes, invent in greenfield an new infra for whole new product offerings and need support + agility to scale, invent in brownfield is to work in parallel with legacy and new through the adoption until legacy deprecation.

People change includes assess, plan, deploy, optimize.

Search This Blog

Notes

Google Cloud Product Fundamentals & Managing Change when Moving to Google Cloud

Comments

Post a Comment