The conversion ratios, the license included comparison, and keeping entitlements compliant once they run in the cloud. Three knowledge checks along the way, and 1 clip from a senior cloud advisor.
This is a taught session, not a talking head. The instructor works through analyst grade slides, and three times the video stops on a question with four options on screen. Pause, commit to an answer, and the next slide explains which option is right and why each of the others is wrong. Once in the session the frame splits and a senior cloud advisor gives the view from inside real Oracle negotiations, and the instructor picks the clip apart when the slides return.
The full narration of this session, section by section, for reading and reference. Guest analyst clips are marked.
Welcome back, session eight of thirty. Two sessions ago you met the most expensive setting in OCI, the license toggle, and I promised we'd come back and do it properly. This is that session. BYOL, bring your own license: the mechanism that lets the perpetual licenses you already own change what OCI costs, more than any discount you will ever negotiate. And I want to frame it the way an analyst should: your license shelf, the thing the Mastery course taught you to count and defend, is also the biggest coupon Oracle has ever issued, worth roughly half the database service rate, forever, if, and it's a real if, you run the bookkeeping properly. Because BYOL is also how estates that were clean on premises acquire brand new audit findings in the cloud. The lever and the liability, same session. Let's go.
Five takeaways. One, the lever: why BYOL moves the OCI price more than negotiated discounts do, and what your shelf is actually worth converted into cloud capacity. Two, the mechanics: which entitlements qualify, the conversion ratios, and the support rule that the whole construct legally rides on. Three, the comparison: BYOL versus license included, modeled honestly, and honestly means three columns, because the two column version that most finance models use hides a six figure annuity. Four, the compliance: BYOL creates a counting obligation that nobody briefs the cloud team about, one license, one place, and we'll make that concrete. And five, the portfolio: BYOL decisions ripple into support restructuring, ULA certification, and your five year platform intent, so we'll zoom out before we finish.
Four numbers to frame it. Roughly two x: the gap between license included and BYOL rates on the database services. The with license rate carries a built in license premium; the BYOL rate strips it out because you're supplying the license. That's the whole trade, renting versus bringing, and on database heavy bills it's the largest rate movement available from any single decision. Two OCPUs: what one EE processor license entitles under BYOL, which in the current unit is about eight ECPUs, session six's conversion doing real work. And pause on the irony, because it's instructive: this is the same EE processor license that the partitioning policy counts so hostilely on VMware. On premises, hostile counting. Inside Oracle's cloud, generous counting. Same license, same vendor, opposite arithmetic, and none of it is an accident, it's a migration incentive wearing a rate card. Active: the support status your licenses must hold. BYOL rides on maintained licenses, and if support lapses, the BYOL right lapses with it, quietly, which is why support status belongs in the quarterly audit. And one place: where each license can be at a time. BYOL is a reallocation, not a duplication, and that single sentence is the entire compliance chapter of this session. The strategy in one line: Oracle deliberately made your old licenses worth more inside OCI than in your datacenter. Use the asymmetry. Audit proof the bookkeeping.
The mechanics, five rules. Rule one, what qualifies: full use perpetual licenses with active support. Database Enterprise Edition, Standard Edition 2, the options, middleware. What does not travel: restricted grants, the application specific and embedded licenses from the Mastery course, their restrictions follow them everywhere, including to the cloud. Rule two, the ratios: one EE processor license covers two OCPUs on the database services, about eight ECPUs. SE2 converts at its own published ratio, NUP entitlements have theirs, and the authoritative source is always the specific service's BYOL table, not a slide, not a memory, the table. Rule three, and this one gets missed constantly: options ride along. A BYOL database using Partitioning, RAC, or the management packs needs those option licenses brought too, at matching counts. Session three of the Mastery course does not get waived by the cloud; the feature usage views exist in cloud databases and they still record everything. Rule four, support stays on: BYOL requires current support on the brought licenses, so the twenty two percent annuity keeps running. Hold that thought, because it's the third column of the model we build in a few minutes, and it's also what makes next session's Support Rewards arithmetic so interesting. And rule five, the toggle executes it: per instance, BYOL or license included, and the contract right means nothing if the console says otherwise, in either direction, paying license included while owning entitlements, or claiming BYOL without them.
First check, all the rules at once. A team wants to BYOL eight EE processor licenses to run a sixteen OCPU database service. The on premises databases run Partitioning, the cloud one will too. Support lapsed on four of the eight licenses last year. What actually works? A, all sixteen OCPUs on BYOL, the licenses exist, support is paperwork. B, only the four supported licenses qualify, covering eight OCPUs, and Partitioning entitlements have to come with them, while the other eight OCPUs run license included or wait for the reinstatement math. C, nothing works, lapsed support anywhere disqualifies the estate. Or D, all sixteen, but Partitioning is free in the cloud. Pause here. Three rules: active support, options ride along, ratios.
The answer is B, and it exercises every rule on the slide. The four licenses with active support qualify, and at two OCPUs each they cover eight of the sixteen. Because the cloud database uses Partitioning, matching Partitioning entitlements travel with them, the options rule, and if the team forgot they owned Partitioning licenses, that's this afternoon's homework. The four lapsed licenses confer nothing until reinstated, and before anyone proposes reinstating, run the Mastery course math: roughly one hundred fifty percent of the lapsed support years plus repricing, which almost always makes license included on the remaining eight OCPUs the cheaper way to close the gap. And the two extreme answers teach the boundaries: lapsed support on some licenses never poisons the others, every entitlement stands on its own paper, and nothing about the cloud makes Partitioning free, the packs and options bill in OCI exactly as stubbornly as they did in your datacenter. Half BYOL, half license included, on one service, is a perfectly normal outcome, and the ledger we build later is what keeps it defensible.
Now the comparison, and the reason I keep saying three columns. Column one, the OCI service rate: BYOL roughly half, license included carries the premium. If the model stops here, BYOL wins every time, and this is where most models stop. Column two, the support annuity, the one nobody puts in the spreadsheet: BYOL requires keeping support on the brought licenses, twenty two percent of the license base, every year, forever. License included needs no such thing, which means for those workloads the support on the equivalent shelf licenses becomes, potentially, terminable, with all the license set caveats the Mastery course taught. Column three, Support Rewards, the claw back: your OCI consumption earns twenty five or thirty three cents per dollar against that same support bill, so part of column two comes back, if consumption is high enough. And there are two structural rows underneath: entitlement freedom, BYOL parks your licenses in the cloud where nothing else can use them, license included leaves the shelf free for on premises needs, ULA math, or support leverage. And exit posture, my favorite underrated row: leave OCI on BYOL and your licenses come home intact, an asset preserved. Leave on license included and you own nothing, the premium bought no asset. The verdict: BYOL usually wins when you own suitable supported licenses. It stops being automatic the moment reinstatement, ULA certification, or a support termination plan enters the picture. Here's our advisor on the version of this that goes wrong most expensively.
Guest analyst The most common finding I make in OCI cost reviews has a boring name: the double payment. It comes in two flavors. Flavor one, the client owns a shelf of EE licenses, pays full support on all of them, and runs their cloud databases on license included rates, because the migration team never knew the shelf existed. They are paying the license premium in the cloud and the support annuity on premises, for the same capability, twice. I found one estate paying both sides of that trade to the tune of four hundred thousand dollars a year, and the fix was a toggle and a ledger. Flavor two is the reverse and it is nastier: the team flips instances to BYOL because the rate is better, without checking the shelf. The entitlements either do not exist, are the wrong edition, or are still covering production servers in the datacenter. That one is not a cost problem, it is an audit finding accruing interest, and Oracle's own telemetry sees the toggle. Both flavors have the same root cause: the people who run the cloud and the people who own the licenses have never been in the same meeting. So that is my actual recommendation. Not a tool. A standing meeting, quarterly, one hour: cloud operations brings the toggle report, licensing brings the entitlement ledger, and they reconcile. Every estate that does this finds money the first time. Every estate that does not is in one of my two flavors right now, and most do not know which.
Two flavors of double payment: renting what you own, or claiming what you don't. Same root cause, the toggle people and the entitlement people never meet. One quarterly hour reconciling the toggle report against the license ledger, and you're immune to both. That meeting goes on the discipline slide at the end, and it should go in your calendar this week.
Check two, the three column model with numbers. An estate BYOLs twenty EE licenses to OCI, keeps paying two hundred ten thousand a year in support on them, as BYOL requires, and the finance model claims the full two x rate saving versus license included. What did the model miss? A, nothing, the rate saving is the saving. B, the two hundred ten thousand support annuity that BYOL requires: the honest saving is the rate gap minus the support kept, partially clawed back by Support Rewards if consumption earns it. C, BYOL rates expire after year one. Or D, license included would have been free with the commitment. Pause here. What does BYOL oblige you to keep paying?
The answer is B. The two column model, rate versus rate, shows a saving of the full gap. The three column model tells the truth: BYOL's low rate, plus the two hundred ten thousand of support the entitlements must keep, minus whatever Support Rewards claws back against that same bill, twenty five cents per consumed dollar, which at healthy consumption is a substantial offset, and next session we compute it exactly. Run it with real numbers and BYOL still wins for most owned estates, but the margin is thinner than the rate gap suggested, and here's when it genuinely flips: when the alternative was terminating that support entirely. An estate planning to shed those license sets, take the license included rate, and kill the annuity can come out ahead, because column two goes to zero on the other side of the ledger. That's not an exotic scenario, it's every estate mid divestment from on premises Oracle. C is fiction, BYOL rates don't expire, and D confuses the commitment discount with the license premium, they stack, they don't substitute. Three columns. Every time. The spreadsheet takes ten more minutes and stops six figure mistakes in both directions.
Now the compliance chapter, the counting obligation nobody briefs the cloud team on. One, the headline rule: one license, one place. The moment a license is allocated to cover an OCI instance, it stops covering whatever it covered in the datacenter. If that on premises database keeps running, it's running unlicensed, and that overlap is one of the most reliable audit findings of the cloud era, precisely because nobody thinks of a migration as a licensing event. Two, the ledger: a dated allocation record, which entitlements, from which ordering documents, cover which cloud instances at which counts. Understand what the console gives you and what it doesn't: it shows toggles, it has no idea what you own. Only your ledger connects the claim to the paper. Three, scaling counts: scale a BYOL service from eight to sixteen OCPUs and your entitlement requirement just doubled, live, in an afternoon, without a purchase order. Elasticity is a licensing event; put an alert on it. Four, options parity: features enabled in the cloud must match options owned, per instance, and I'll say it a third time because it keeps being true, the feature usage views record everything, in the cloud too. And five, the audit posture: a BYOL estate gets asked exactly two questions. Show me the entitlements. Show me they're not counted twice. The ledger answers both in an afternoon. Without it, the audit answers them for you, over six months, with a settlement at the end.
Zoom out, because BYOL is a portfolio move disguised as an instance setting. Interaction one, support restructuring: licenses parked in BYOL must keep support, which means their license sets are locked out of any termination plan for as long as they're parked. If the estate has a support reduction strategy, and after the Mastery course it should at least have considered one, the sequencing matters: decide the support endgame first, then allocate BYOL only from the sets you intend to keep forever. Allocating first and planning later welds the annuity in place. Interaction two, ULAs: inside a ULA term, you generally don't BYOL your ULA products, cloud deployments count toward certification under the averaging rules the Mastery course covered, and BYOL from those licenses only makes sense after exit, from the certified counts. The ULA timeline dictates the BYOL timeline, never the reverse, and getting that backwards can cost you certification volume exactly when it matters most. And interaction three, the endgame test: before parking licenses in the cloud, answer honestly where the estate is in five years. All in on OCI: BYOL freely, the parked licenses are working assets. Multi cloud, or shrinking Oracle: keep the shelf flexible, because license included, at its premium, is buying you freedom, and freedom is sometimes worth the rent. The good news: a wrong BYOL allocation is reversible, licenses can come home. But every quarter it stands, it shapes the support bill, the audit surface, and your negotiating posture, so it's a decision to renew annually, not a default to forget.
Last check, the migration overlap. A company BYOLs its EE licenses to OCI, but keeps the on premises databases those licenses covered running during a nine month migration overlap, for safety. The accurate read: A, fine, migrations get a grace period automatically. B, an unlicensed gap: the same licenses cannot cover both ends of the migration, so the overlap needs separate cover, planned as a cost of the migration, or a negotiated bridge from Oracle. C, fine as long as the on premises databases are only used for fallback. Or D, a problem only if Oracle audits during those nine months. Pause here. One license, one place. Where are the licenses?
The answer is B. The licenses are in the cloud, the datacenter databases they used to cover are still running, and running unlicensed, for nine months. There is no automatic migration grace period, that's answer A's wishful thinking, and C's fallback theory collides with counting rules you already know: a fallback database is an installed, running database, and installed counts. The Mastery course's ten day failover rule covers one passive node in a cluster, not a nine month parallel estate. So the clean plays, and there are two: run the overlap on license included rates and flip the toggle to BYOL at cutover, pricing the overlap as what it is, a cost of the migration; or negotiate an explicit migration bridge into the deal up front, dual use rights for a defined window, which Oracle grants readily for migrations it wants, and this is a migration into their cloud, they want it. Answer D deserves its own sentence: the audit lottery is not a compliance position. The gap exists whether or not anyone inspects it, the configuration records and feature views date everything, and a nine month overlap discovered two years later is exactly the kind of finding that settlements are built around. Plan the overlap. It's one line in the deal.
The BYOL discipline, five habits, and you'll recognize the shape by now: owned, dated, repeated. One, ledger first: no BYOL toggle flips until the allocation is written down, entitlement, source order, target instance, count, date. The ledger precedes the toggle, always, because reconstructing allocations after the fact is archaeology. Two, the quarterly toggle audit: session six's habit, upgraded with entitlement eyes. Every database service, BYOL versus license included, checked against intent and against the ledger. This is the advisor's standing meeting, cloud operations and licensing in one room, one hour, and every estate finds money the first time. Three, source decommission tracked: every BYOL allocation carries the date its on premises source was retired, or an explicit line covering the overlap. That column is your audit answer for the double counting question. Four, scale alerts: elasticity events on BYOL instances notify the license owner, because scaling consumes entitlements as surely as it consumes credits, and only one of those shows up on an invoice. And five, the annual portfolio review: the whole BYOL map held against the support strategy, the ULA timeline, and the five year platform intent. Parked licenses are a decision renewed annually, not a default inherited forever. Five habits, maybe three hours a quarter, and the biggest discount in OCI stays a discount instead of becoming a finding.
Session eight, three sentences. One: BYOL is the biggest single lever on the OCI price, roughly half the database service rate, riding on actively supported full use licenses at two OCPUs per EE processor, with the options brought alongside, because the cloud waives none of the options rules. Two: the honest model has three columns, the rate gap, the support annuity BYOL keeps alive, and the Support Rewards claw back, and for estates shedding support, license included sometimes genuinely wins, so run the columns before the toggle. Three: one license, one place, and the ledger, the quarterly reconciliation, and the decommission dates are what make the discount audit proof instead of audit bait. Next session completes the module two money story: Support Rewards, the loop that connects your OCI consumption back to the support bill BYOL just obliged you to keep. The twenty five and thirty three cent accrual in full, a worked example that zeroes a real support bill, and how to spend the loop at a renewal. It's the session where three different threads of this course finally pay each other off. See you there.
Homework, about an hour, and it builds the ledger this session kept invoking. One, map the toggles: every database service instance, BYOL or license included, from the console. You did this in session six; today it gets its second column. Two, attach entitlements: for each BYOL instance, which licenses, from which ordering document, at what count. Every blank in that column is a finding, yours to fix quietly or an auditor's to price loudly. Three, check the sources: for each allocated license, is the on premises workload it used to cover actually retired? Date it, or flag the overlap and decide its cover this quarter. Four, verify support status on every allocated set, because a lapsed set under a BYOL instance is a silent disqualification that nothing in the console will ever flag. And five, price one comparison: take your largest license included instance and remodel it as BYOL with the three column math, rate gap, support kept, Rewards clawed back. If BYOL wins, and if you own the shelf it usually does, you've just found next quarter's saving, and it's a toggle, not a project. An hour of work for possibly the best hourly rate in this course.
Five reads before next session, all free on redress compliance dot com. First, the Oracle BYOL comprehensive guide, the full rulebook behind today's mechanics, including the service by service tables. Second, BYOL versus license included on OCI and Exadata Cloud at Customer, today's three column model worked across the service tiers, with the Exadata wrinkle module three will need. Third, the BYOL cloud transfer and license retention rules, what happens to entitlements as they move in and out of clouds, the fine print behind licenses come home. Fourth, the OCI licensing and cost guide, the rate mechanics all of this plugs into. And fifth, the Support Rewards guide, tomorrow's whole topic and today's third column, read it and next session's arithmetic will land twice as hard. That's session eight. The shelf is a coupon, the model has three columns, and the ledger keeps it all honest. See you at session nine.